Search

Grinding is one of the most widely employed machining processes in manufacturing. Achieving a successful grinding process characterized by low fault rates and short cycle times can significantly improve overall productivity and process efficiency. Nonetheless, finding optimal values for grinding process parameters is challenging due to complex underlying dynamics. Therefore, this work proposes a data-driven system that exploits various machine learning techniques and metaheuristic optimization algorithms to optimize grinding process parameters. Using data collected from grinding processes, the proposed system constructs a machine learning-based fault detection model and employs that model to define variable range constraints. In addition, a Gaussian process-based cycle time estimation model is developed. Process parameter optimization is performed using various metaheuristic algorithms based on the aforementioned methods. Experiments with actual internal cylindrical grinding process data have proven the proposed system's effectiveness during process parameter optimization. Furthermore, real-world validation data verifies the final optimization solution, reducing the fault rate and process cycle time by 77.83% and 17.64%, respectively. In-depth interviews with six domain experts in the grinding process also verify the proposed system's validity and real-world applicability. The proposed data-driven system is expected to bring substantial improvements in process productivity, especially when applied to manufacturing sites in practice. [ABSTRACT FROM AUTHOR]

Machine learning classifiers have been widely used for crop mapping; however, employing numerous features can reduce classifier accuracy. Feature selection is a widely recognized technique that addresses the issue of high dimensionality by choosing the most relevant subset of features with the highest degree of relevance and practicality. Therefore, the optimal combination of feature selection methods and classifiers for high-precision crop mapping remains an open problem. In this study, we introduce a novel hybrid feature selection approach to obtain the optimal subset of features for effective crop mapping. Initially, spectro-temporal remote sensing features from the dataset were ranked using two distinct feature selection techniques: Mutual Information (MI) and ReliefF filter. The most relevant features from each filter-based approach were combined into a union subset. Subsequently, the Gray Wolf Optimization (GWO), a metaheuristic optimization technique, was applied to enhance the feature set generated in the initial step. In the final stage, a random forest classifier leveraged the optimized feature subset for accurate crop type prediction. The effectiveness of the proposed approach was evaluated in Behiera governorate, Egypt. Comparative analysis against existing crop mapping methods demonstrated the superior performance of the proposed approach, achieving an accuracy rate of 82%. Furthermore, this approach showcased a significant reduction in the feature count, streamlining the set from 308 down to only 26 features. [ABSTRACT FROM AUTHOR]

Twin screw granulation (TSG) has recently been emerged as a novel approach for the continuous wet granulation of fine particles (i.e., powders) in the pharmaceutical industry. The presence of brilliant advantages like the ability of operation at very low liquid concentrations and excellent product consistency has made this technique promising. Except positive points, the existence of major challenges like scalability and flexibility in the processing regimes has enhanced the importance of deeper investigations towards true recognition of this process. The central aim of this theoretical article is to develop the modeling process of TSG employing four machine learning models and one metaheuristic algorithms in a hybrid approach. Screw speeds, material throughputs, liquid binder (water)-to-solid ratios, and screw configurations are known as important parameters of TSG process, which were validated via their comparison with the obtained experimental data. GBR, SGD, and SVR were finally selected for 3 targets with their best combinations of hyper-parameters employing FA. The output is based on d-values (d10, d50, d90) for the granulate particle size distribution (PSD). Final models have R2 scores of 0.919, 0.960, and 0.877 for d10, d50, d90 outputs, respectively. [ABSTRACT FROM AUTHOR]

Industry 4.0 has brought modernization to the production system through the network integration of the constituent entities which, combined with the evolution of information technology, has enabled an increase in productivity, product quality, optimization of production costs, and product customization to customer needs. Despite the complexity of human thought, artificial intelligence tries to replicate it in algorithms, creating models capable of processing databases with a high volume of information, and generating valuable information for decision making. Within this area, there are subfields, such as Machine Learning and Deep Learning, which, through mathematical models, define patterns to predict output data from known input data. In addition to this type of algorithm, there are metaheuristic models capable of optimizing the parameters required in Machine Learning and Deep Learning algorithms. These intelligent systems have applications in various areas such as industry, construction, health, logistics processes, and maintenance management, among others. This paper focuses on Artificial Intelligence models addressing Industry 4.0 approach. [ABSTRACT FROM AUTHOR]

The mixed data sampling (MIDAS) model has attracted increasing attention due to its outstanding performance in dealing with mixed frequency data. However, most MIDAS model extension studies are based on statistical methods or machine learning models, which suffer from insufficient prediction performance and stability in small sample environments. To solve this problem, this paper proposes a novel mixed frequency sampling discrete grey model (MDGM(1, N)), which is a coupled form of the MIDAS model and discrete grey multivariate model. By adjusting the structure parameters, the model can be adapted to different sampling frequencies data, and degenerate into several types of grey models. Then, the unbiasedness and stability of the model are proved using the mathematical analysis method and numerical random experiment. The meta-heuristic algorithm is introduced to obtain the optimal weight parameters and the maximum lag order, improving the model's fitting ability to mixed frequency data. To demonstrate the effectiveness of the new model, a model evaluation system consisting of traditional evaluation metrics and a monotonicity test is established. Taking four hard disk drive failure datasets as research cases, the performance of the proposed model is compared with seven mainstream benchmark models. The results show that the proposed model has excellent applicability and outperforms other competition models in terms of validity, stability, and robustness. Furthermore, it is observed that the reported uncorrectable errors and the command timeout have a greater impact on hard disk drive failure. Finally, the new model is employed to forecast the failure of four hard disk drives. The forecasting results indicate that in the next four time points with a cycle of 21 days beginning in April 2023, the failure of the smaller capacity hard disk drives (0055 and 0086, corresponding to 8TB and 10TB) show a decreasing trend, reaching 67.442% and 89.7683%, respectively. The failure of the other larger capacity hard disk drives (0007 and 0138, corresponding to 12TB and 14TB) has increased, with a growth rate of 17.1016% and 123.7899%. [Display omitted] • A novel mixed frequency sampling discrete grey model is proposed. • The proposed model is proved to be unbiased and stable theoretically. • A Chimpanzee Optimization algorithm is introduced to globally search for weight parameters and maximum lag order. • A new model evaluation system for evaluating the effectiveness of prediction models is constructed. • The proposed model outperforms the other seven benchmark models in four case studies. [ABSTRACT FROM AUTHOR]

Gender identification from video is an emerging research field that aims to automatically classify the gender of individuals based on video data. Due to the numerous applications for this task, it has received a lot of attention, including surveillance, human-computer interaction, and targeted marketing. In this study, we propose a gender identification system that utilizes the Pelican optimizer algorithm in combination with a Support Vector Machine (SVM) classifier. The Pelican optimizer is a metaheuristic algorithm inspired by the hunting behaviour of pelicans and has shown promising results in solving optimization problems. Pelican optimizer algorithm (POA) is applied to optimize the SVM parameter selection process such as kernel function. The POA algorithm searches for an optimal subset of parameters that maximizes the classification performance of the SVM model after the application of preprocessing and feature extraction techniques such as Local Binary Pattern (LBP). Finally, the selected optimized parameters otherwise known as POA-SVM classifier learns a decision boundary based on the labeled training data. The POA-SVM model is trained to distinguish between male and female samples and generalize the classification to unseen video data. Experimental evaluations are conducted using a benchmark dataset consisting of video samples with labelled gender information. The effectiveness of the suggested system is contrasted with other cutting-edge gender identification techniques. The results demonstrate the effectiveness of the Pelican Optimization Algorithm-SVM system, showing improved accuracy of 95%, and sensitivity of 94.4% at a faster recognition rate in gender classification from video data. [ABSTRACT FROM AUTHOR]

There are various paradigms of learning. One such paradigm is learning from knowledge assimilation. In machine learning, such a paradigm may be considered where concepts (classification models) generated from different algorithms applied to the same training data are combined to come up with a single integrated improved concept (classification model). This is different from combining outputs of various models as in Bagging or Boosting. Thus, learning from knowledge assimilation creates different alternative models that work on different parts of the search space. Such models can be separately optimized so that the best-performing model can be selected for the given problem. For optimization there are several algorithms and Chemical Reaction Optimization (CRO) is a chemistry-based nature-inspired optimization algorithm. In this algorithm, a set of optimization approaches are introduced corresponding to various types of chemical reactions. Classification models in our approach are optimized using a technique called Elitist CRO(ECRO). ECRO is an improvement over CRO, where the optimization approaches are modified making the convergence of the algorithm better than CRO. Although for classification CRO models based on neural networks exist, however, they suffer from interpretability. This paper proposes an application of a new Hybrid Elitist Chemical Reaction Optimization (HECRO) algorithm to generate interpretable classification rules. The proposed algorithm is an extension of CRO and ECRO, in the manner of region exploration of the search space, which is expanded by combining classification rules generated from various algorithms as the initial population following learning from knowledge assimilation. Experimental results with 20 datasets drawn from the UCI Machine Learning data repository show improvement over several performance measures in comparison to the base classifiers used for population generation in HECRO. [ABSTRACT FROM AUTHOR]

Smart manufacturing involves the use of a variety of automation solutions, such as robotics, machines with embedded software, and advanced sensors collecting vast quantities of data. Efficient control of a complex composition of such solutions, as well as the analysis of the collected data, are essential for improving the efficiency of the production processes and decision-making. Data analysis and process optimization are enabled through the application of state-of-the-art optimization and machine learning algorithms. However, the efficient use of these algorithms often depends on the careful selection of the parameters, which on its own is a process that requires a high degree of expertise and time. Therefore, another class of algorithms can be applied that is designed to discover the optimal parameter configuration given the specific nature of the manufacturing process and the used algorithm. In this work, we systematically analyze the published literature to discover which parameter selection techniques are used in the context of Industry 4.0, for which processes, and how these benefit from automated parameter selection. Within our literature review, we discover nine relevant publications, most of which concentrate on parameter selection for machine learning algorithms through various numerical optimization and metaheuristic techniques. [ABSTRACT FROM AUTHOR]

Our article aims to resolve the scheduling instances in a job shop, hereinafter referred to as JSSP. The objective is to minimize the makespan, or the maximum amount of time needed to complete all tasks. In this contribution, we study an improved hybrid metaheuristic for solving active scheduling. This hybrid metaheuristic improves the outcome based on the previous output concerning some policies. The space solution is investigated using the local search algorithm by diversifying the neighborhood structure. The new matrix heuristic is used to design an admissible schedule and evaluate the makespan of the neighbor of the present solution. Benchmark instances are simulated using MATLAB to evaluate the efficiency of our metaheuristic. The new metaheuristic is quick to compute, straightforward, and simple to use. Comparative studies between different neighborhoods used by our metaheuristic, emphasize that using the hybrid's combination of neighborhoods gives the best results. The results' simulations demonstrate the quality of the suggested metaheuristic. [ABSTRACT FROM AUTHOR]

Electricity is one of the most important energy sources for the progress of human society. Photovoltaic power generation (PPG) has great potential as a renewable energy source to support sustainable development. However, the unpredictability of photovoltaics because of the uncertain weather conditions, time lags and the complex interactions among related factors may result in instabilities in PPG, which limits its applications in the electric power system. Therefore, accurately identifying the related factors and representing the interactions are the key issues in predicting PPG. For this purpose, a novel seasonal grey prediction model with time-lag and interactive effects, denoted as SGMLI(1, n), is developed to predict the capacity of photovoltaic power generation. More specifically, a time-lag driving term, a periodic driving term, and an interaction driving term are proposed to represent the lag relationship, periodicity, and the interactive effects between two factors, respectively. Utilizing the least square method and the Whale Optimization Algorithm (WOA), the optimal solution to the nonlinear programming problem is estimated. For elaboration and verification, six grey forecasting models, two statistical techniques, and two machine learning models, are employed for comparison with the SGMLI(1, n) model. Experimental results reveal that the SGMLI(1,n) outperforms benchmark models in forecasting PPG. • A seasonal grey model with with time-lag and interactive effects is developed. • The new model can deal with the prediction under the interactions of related factors. • The novel model adapts to the time series characterized by time delay and periodicity. • The new method improves the applicability of the grey forecasting model. • Empirical results demonstrate the novel model's excellent forecasting performance. [ABSTRACT FROM AUTHOR]

WSN (Wireless Sensor Network) is considered as one of the promising technologies which are utilized in various fields for various applications. Along with it, MANET (Mobile Ad Hoc Network) has attracted a distinct attention, as they serve communication means across innumerable domain. In recent years, due to the development of both wired and wireless technologies, attacks are becoming more frequent and these attacks can be in various form such as worm hole, black hole, grey-hole, sinkhole and others. These attacks can cause loss of security, increase the number of drop packets, decrease packet delivery ratio which result in poor routing performance. In order to overcome these issues, the proposed study employs PBCS (Particle Bee Colony Swarm) algorithm for finding the shortest path between the nodes, which helps in reducing the routing cost and makes the model more efficient and effective. In addition to this, Hybrid AdaBoost-Random forest algorithm is utilized in proposed model. This model helps in reducing the training time which makes the model more reliable and efficient than existing models. It can be implemented in huge dataset and in Hybrid AdaBoost-Random Forest algorithm, estimator of random forest is used, which result in high accuracy and low loss. AODV (Ad-Hoc on Demand Vector) protocol helps to identify the prevention of attack as this routing protocol perform better for longer and extensive duration of traffic than other protocols. The performance of the proposed model is evaluated using different metrics which includes accuracy, recall, precision, sensitivity and specificity. Besides, energy consumption, throughput, network lifetime, delay of the proposed model is also evaluated. The proposed model is then compared with various existing models in order to determine the efficiency and effectiveness of the proposed framework. [ABSTRACT FROM AUTHOR]

• Innovative AI-based approach to designing an acoustofluidic micromixer is proposed. • Response Surface Methodology is used to mathematically model the system. • Metaheuristic algorithms and RSM are employed to for data-driven optimization. • DE outperformed NSGA-II in finding optimal solutions with better functionalities. • Number of sharp edges significantly influences mixing efficiency in high voltages. Micro-mixing is an essential step in various biological and chemical analyses and synthesis processes. In the present study, an acoustofluidic micromixer is designed, modeled, and its functionality has been investigated. Intelligent acoustofluidic is proposed as an evaluation-optimization method, based on Machine Learning algorithms. The first step involves designing a parametric geometry with multiple sharp edges. Following that, mathematical modeling of the system is carried out using Response Surface Methodology (RSM) and computational fluid dynamics (CFD). Geometrical parameters including the sharp edge tip angle and the channel thickness, along with the peak-to-peak actuation voltage were investigated. It was revealed that, due to the inherent nonlinearity of acoustic streaming, each design parameter could significantly impact the flow patterns and mixing performance of the acoustofluidic mixer. The mathematical surrogate model obtained from data-based modeling procedure is then employed for the artificial intelligence multi-objective optimization algorithms, i.e. Differential Evolution (DE) and Non-dominated Sorting Genetic Algorithm two (NSGA-II). The two objective functions are the mixing index and the mechanical energy consumption index (ME). The derived optimum solutions had low Mehcanical Energy (ME) values while yielding homogeneity in a wide range of applied voltages. This ensures the controllability of the proposed system, which is an invaluable feature in synthesis processes to determine the critical physical properties of the nanoparticles. Through the algorithm evaluation process, which was conducted to distinguish the more accurate and robust optimization method, it was revealed that the DE algorithm outperformed NSGA-II in generating optimum solutions with superior functionalities. In the final section, three optimum models, operating under three different V p - p values, 10, 20, 30, were selected, and their acoustofluidic characteristics were analyzed. It was concluded that utilizing baffles with three sharp edges at low V p - p values results in a 96% Mixing Index and 48% decrease in the power consumption of the system. However, at high V p - p values, baffles with one sharp edge demonstrated superior mixing functionality, with a MI of 98%. Moreover, it is revealed that utilizing baffles with three-sharp edges results in an 80% increase of Mixing Index of the acoustofluidic micromixer. [ABSTRACT FROM AUTHOR]

• A novel integrated framework for the diagnosis of breast cancer patients is proposed. • A scheme combining feature selection and hyperparameter optimization is proposed to improve the prediction accuracy of the model. • Compared with the existing models, the proposed model has higher performance and generalization ability. In this paper, a new integrated scheme is proposed to accurately predict breast cancer, help doctors make early diagnosis and treatment plans, and improve the prognosis of patients. We selects five mainstream machine learning models: support vector machine (SVM), artificial neural network (ANN), random forest (RF), extreme gradient boosting (XGBoost) and adaptive boosting (AdaBoost). The Wisconsin Breast Cancer Database (WBCD) and Wisconsin Diagnostic Breast Cancer (WDBC) are used as datasets to investigate the predictive performance of these single and ensemble models. Then, we use multiple linear regression method for feature selection (FS), the experimental results show that the change of feature subset will significantly affect the performance of the model. The recall and f1-score of the five models are improved by 1.19% and 0.84% on average. After that, we apply whale optimization algorithm (WOA) to optimize the hyperparameters of the model to improve their prediction performance. In the best-case scenario, the model demonstrated improvements of 1.02% in accuracy and 1.82% in precision. In addition, we ensemble these models by stacking, investigate the performance changes of the ensemble model when different models are used as meta learners. Finally, the FS-WOA-Stacking model achieves 99.56% accuracy on WBCD and 99.65% accuracy on WDBC. Compared with the existing breast cancer prediction models, the performance of the proposed model is at an excellent level. [ABSTRACT FROM AUTHOR]

[Display omitted] • A machine learning model was proposed to predict the tensile properties of 2219-T8 Al alloy TIG welded joint. • Relationship between tensile strength of joint and weld geometry, weak zone properties was developed. • The collaborative regulation mechanism of geometry and property was discussed. • The machine learning model could be used to optimize and guide welding process. The tensile properties of 2219-T8 aluminum alloy TIG welding joint were significantly affected by the microstructure, local mechanical properties and weld geometry. This paper proposed a machine learning model to predict and optimize the tensile properties of 2219-T8 aluminum alloy TIG welding joint. The relationship between tensile strength of joint and weld geometry, weld zone and partially melted zone (PMZ) properties was developed by Kriging model combining whale optimization algorithm (WOA). This surrogate model demonstrated a high precision, with R2 = 0.952 and RMSE=3.77 MPa. The surrogate model, which also served as a welding process guide, was utilized to determine the ideal weld geometry corresponding to various weak zone properties. By applying the optimization process based on the surrogate model, the optimized joint strength coefficient reached 70 %, and elongation exceeded 4 %. The collaborative regulation mechanism of geometry and property was also discussed. [ABSTRACT FROM AUTHOR]

• Sustainable Development Goal 7 was addressed in the off-grid multi-region energy study. • A multi-region design load system was proposed to reduce the total annual cost. • Several hybrid approaches were derived for the modelling of energy demand. • Hybrid ML algorithms were utilized and compared with the traditional ML method. • The outcomes depicted the reliable accuracy of hybrid models. Reliable energy demand estimation and optimal sizing of a stand-alone photovoltaic/wind/battery hybrid energy system are critical for achieving sustainable development goals. This study presents a hybridized design of an off-grid multi-region energy system to reduce the total annual cost of the system based on a genetic algorithm. A stand-alone kernel support vector regression was used to predict the short-term load demand variability of four major states in Nigeria, namely, Kano, Abuja, Niger, and Lagos. To calibrate the models, primary data was obtained from the four states: load demand (W), temperature (oC), wind speed (m/s), and solar radiation (Wh/m2). The model was evaluated in calibration and verification phases using four key metrics: Nash-Sutcliffe Efficiency, Mean Squared Error, Root Mean Squared Error, and Correlation Coefficient. The model's accuracy was visualized using a newly emerged fan plot and a two-dimensional Taylor diagram. Three kernel support vector regression-C models (C1, C2, and C3) were derived based on the input combination. The outcomes proved that the kernel support vector regression-C3 model tends to perform better in prediction accuracy for all the cities, followed by kernel support vector regression-C2 and-C1. Compared to other cities, Lagos (kernel support vector regression-C3) has the lowest prediction accuracy, while Niger (kernel support vector regression-C3) has the highest prediction accuracy. To enhance the stand-alone approach, hybrid adaptive neuro-fuzzy-based particle swarm optimization, and biogeography-based optimization algorithms were employed. The optimization results revealed that the adaptive neuro-fuzzy-based particle swarm optimization model outperformed the biogeography-based optimization approach in the calibration and verification phases. Niger State demonstrated the highest performance metric values, with Kano, Lagos, and Abuja following closely behind. These findings proved that adaptive neuro-fuzzy-based particle swam optimization is a promising model for load demand forecasting in Nigeria and holds potential for use in other developing countries. [ABSTRACT FROM AUTHOR]

Efficient power management is fundamental for organizations and systems to conserve resources, reduce costs, and improve environmental sustainability. To achieve the mentioned purposes, it is vital to perceive power consumption patterns and trends to make informed decisions about energy usage. By evaluating power consumption, inefficiencies and areas for improvement can be identified. In this regard, this paper presents a comprehensive study to predict power usage in a non-residential building HVAC system. The study employs the integration of machine learning models and optimization algorithms to achieve accurate predictions. The results demonstrate the effectiveness of the approach, with the CatBoost-AO model depicting superior performance in comparison with other models across a range of statistical evaluation metrics including NSE, JSD, RMSE, MAPE, VAF, KLD, as well as R2 and runtime. Remarkably, the CatBoost-AO model achieves the maximum R2 value of 0.91, indicating strong predictive capability. Overall, the presented study highlights the potential of employing machine learning and optimization techniques to improve power management and resource efficiency in HVAC systems, contributing to more sustainable and cost-effective operations in non-residential buildings. • Study conducts power consumption predictions using a comparative approach. • CatBoost and AdaBoost models integrated with optimizers for enhanced performance. • Aquila Optimizer, Arithmetic Optimization Algorithm, and Parallel Particle Swarm Optimization are utilized. • Hybrid models evaluated using metrics like NSE, JSD, RMSE, and R2. • CatBoost-AO model excels, achieving an R2 of 0.91 in testing, outperforming other models. [ABSTRACT FROM AUTHOR]

Optimization problems can be seen in numerous fields of practical studies. One area making waves in the application of optimization methods is data mining in machine learning. An important preprocessing technique of data mining where irrelevant variables are discarded from the datasets and holding onto variables with important information is referred to as feature selection (FS). FS is critical to tackling the 'curse of dimensionality' by reducing the number of features, minimizing computational expensiveness and maximizing the accuracy of the machine learning models. Swarm Intelligence (SI)-based meta-heuristic algorithms (MAs) have been widely employed to solve several optimization problems like FS. However, common drawbacks identified with these algorithms include getting trapped in local optima, especially in situations where the search space is large (high dimensional space). This study proposes a new hybrid SI-based MA called Particle Swarm-guided Bald Eagle Search (PS-BES). The algorithm utilizes the speed of Particle Swarm to guide Bald Eagles in their search to ensure a smooth transition of the algorithm from exploration to exploitation. Additionally, we introduce the Attack-Retreat-Surrender technique, a new local-optima escape technique to enhance the balance between diversification and intensification of PS-BES. To establish the outstanding performance of the proposed algorithm, PS-BES is comprehensively analyzed utilizing 26 Benchmark functions. Further, the practicality of PS-BES is highlighted by its binary version for feature selection and evaluated using 27 classification datasets from the UCI repository. The results prove the overall superiority of PS-BES and bPS-BES as opposed the 10 state-of-the-art algorithms employed in the study. • Metaheuristic methods have proven to effectively tackle optimization problems. • Wrapper-based feature selection technique efficiently evaluate important subspaces. • The Attack-Retreat-Surrender technique fosters the robust search strategy in PS-BES. • PS-BES and bPS-BES show outstanding performance over the competing methods. [ABSTRACT FROM AUTHOR]

Hot compression simulation tests were carried out on a 56Ni–32Ti–12Hf alloy at temperatures ranging from 800 to 1000 °C and strain rates ranging from 0.001 to 1 s−1. The study aimed to explore the hot deformation behavior and microstructure evolution mechanism of the alloy. Two flow stress prediction models were developed: a strain-compensated Arrhenius constitutive (SCAT) model based on phenomenology and a chaotic mapping adaptive inertia weight whale optimization algorithm-optimized back propagation neural network (CIWOA-BPNN) model based on machine learning. In comparison to the SCAT model, the CIWOA-BPNN model demonstrates higher prediction accuracy, improved error correlation coefficient, decreased average relative error, and lower root mean square error and mean absolute error. The hot processing map of the 56Ni–32Ti–12Hf alloy was developed using dynamic material modeling (DMM) theory and flow stress data predicted by the CIWOA-BPNN model. The optimal hot processing range was found to be between temperatures of 875–975 °C and strain rates of 0.001–1 s−1. The study found that within the optimal processing interval, the softening mechanisms observed were discontinuous dynamic recrystallization (DDRX) and continuous dynamic recrystallization (CDRX). An increase in power dissipation efficiency (η) led to an increase in the dynamic recrystallization (DRX) fraction from 27.1 % to 41.9 %, while the substructure fraction decreased from 34.2 % to 19.8 %. The decrease in the number of dislocations around the DDRX grains further validated the accuracy of the identified optimum machining interval in this research. • A new High temperature shape memory alloy 56Ni–32Ti–12Hf as an experimental material. • Chaotic mapping adaptive inertia weight whale optimization algorithm neural network model (CIWOA-BPNN) applied to high temperature flow stress prediction of 56Ni–32Ti–12Hf alloy. • The prediction accuracy of the CIWOA-BPNN model is higher than most reported predictive models. • A hot processing map based on a Machine-learning Approach and Dynamic Materials Model. [ABSTRACT FROM AUTHOR]

With the expansion of power grid, unaffordable computational cost and time will pose serious challenges of time-efficient scheduling in unit commitment problem (UCP). However, existing optimization methods, i.e., mathematical programming methods and meta-heuristic algorithms, are powerless and time-consuming to handle computationally expensive UCP (CEUCP). Thus, reinforcement learning methods with strong inference and time-saving performances are motivated to solve the computationally expensive challenges in tackling CEUCPs. In this paper, a novel expert knowledge data-driven based actor–critic (AC) reinforcement learning methodology is proposed for solving CEUCPs. Specifically, in the proposed AC reinforcement learning methodology, expert knowledge, data-driven surrogate model, and improved meta-heuristic algorithm are integrated for further performance enhancement. Firstly, a novel action selection mechanism (based on the expert knowledge of thermal units characteristic) is integrated into AC to improve the efficiency of network training. Secondly, an improved extreme learning machine (ELM) data-driven surrogate model is proposed to build reward function in AC. In detail, original cost function in reward is replaced by a lightweight ELM model. Shape distance is integrated into ELM for enhancing accuracy. Finally, original marine predators algorithm (MPA) is improved for obtaining optimal dispatching decisions and rewards of AC method quickly and correctly. Original search pattern is replaced by quantum based representation for boosting convergence. The excellent performances of the proposed AC framework are verified by simulations of 10-units, 100-units, and 100-units with wind energy test systems. • Computationally expensive unit commitment (UC) is modeled by Markov process. • Use reinforcement learning (RL) method for obtaining more economical unit status. • Propose an expert knowledge-based action selection method for rapid convergence. • Improve a time-saving extreme learning machine model to build the reward function. • Improve marine predator algorithm to get dispatch decisions quickly and accurately. [ABSTRACT FROM AUTHOR]

• A large-scale solar linear Fresnel system with thermal oil storage tank is experimentally studied. • The maximum instantaneous heat production was found to be 62.3 kW. • The proposed system achieved tank thermal oil temperatures up to 213.2 °C. • relevance vector machine (RVM) was employed to model the system operation. • Simulated annealing algorithm (SAA) was utilized to explore the optimal structure of RVM. • RVM-SSA had a unique accuracy in predicting the performance of the linear Fresnel system. The advancement of solar linear Fresnel reflector (LFR) systems as a viable technology for green heat and power generation has recently garnered significant attention from both researchers and authorities. Regrettably, this technology's lack of experimental performance data and operational parameters hinders its in-depth characterization, complicates modeling and design efforts, and constrains its practical applications. A large utility-scale solar LFR system with evacuated compound receiver is experimentally investigated and numerically modeled in this work. The site selection, technical design, and experimental implementation of the system are technically introduced. Moreover, the experimental performance of system is examined in the harsh outdoor conditions of Al-Khobar, Saudi Arabia and thoroughly assessed in terms of specific heating power per solar collection area, total heat power production, exit hot oil temperature, energy efficiency, and net exergy efficiency. More so, the long-term operation performance of the established LFR system, for the first time, is modeled and predicted using the relevance vector machine (RVM) modeling as a novel machine learning benchmark. Additionally, an innovative version of metaheuristic algorithms, namely simulated annealing algorithm (SAA) is incorporated with the original RVM to explore the optimum RVM hyperparameters towards maximizing the prediction accuracy. Moreover, using eight different statistical measures, the RVM-SAA model is compared with typical RVM and classical ANN. The experimental findings demonstrate the capability of the LFR system to generate approximately 62.30 kW of useful thermal power, with energy, thermohydraulic, and exergy efficiencies reaching up to 35.10 %, 25.95 % and 7.82 %, respectively. Furthermore, the operation of this new LFR system can lead to temperature levels up to 213.25 °C at a continuous thermal oil flow at a rate of 9.0 m3/h. Furthermore, the statistical analyses demonstrated the superiority of the RVM-SAA approach compared to other models investigated for predicting the system performance profiles. Specifically, RVM-SAA achieved the highest deterministic coefficient values of 0.9997 and 0.9988 and the lowest RMSE of 0.4090 and 2.5150 for predicting the system's thermal power production and the outlet oil temperature, respectively. [ABSTRACT FROM AUTHOR]

Alzheimer's disease (AD) is a chronic disorder characterized by progressive cognitive dysfunctions and memory loss. Electroencephalography (EEG) is a non-invasive tool to detect AD using machine learning models and signal-derived features. Feature selection enhances AD detection performance, reduces training and testing time, and builds simpler models. Metaheuristic algorithms (MHAs) have successfully selected optimum features in several classification tasks and can be valuable in enhancing AD detection. In this work, an AD detection model is proposed based on a systematic investigation of adaptive signal decomposition techniques and MHAs. The four techniques comprising empirical mode decomposition, variational mode decomposition, discrete wavelet transform, and Low-Complexity Orthogonal Wavelet Filter Banks (LCOWFBs) decompose EEG signals into subbands. Thirty-four features based on signal, temporal, spectral, and entropy analysis are extracted for each subband. The salient features are selected using seven different MHAs: particle swarm algorithm, salp swarm algorithm, reptile search (RSA), Grey Wolf, dwarf mongoose optimization, snake optimizer, and Fick's law algorithm (FLA). The two publicly available AD EEG datasets validated the performance of the proposed framework. LCOWFBs achieved the highest binary classification accuracy of 99.72% using 15 salient features selected by RSA and multi-class accuracy of 88.92% with seven salient features selected by FLA. The present method investigates brain regions that are affected by AD. The results indicate that the combination of LCOWFBs and RSA achieved maximum AD detection performance for both datasets with shorter computational time than earlier reported methods in the literature. The present model can be extended to identify chronic disorders, including hypertension, sleep disorders, and Parkinson's disease. • Developing an Metaheuristic Algorithms-based optimal feature selection model for the detection of AD and MCI from normal. • Presenting the signal processing decomposition techniques EMD, VMD, DWT, and LCOWFBs for feature extraction in bandwidth-duration domain. • Computing temporal, spectral, and entropy-based thirty-five features to explore the significant brain regions. • Evaluating classification performance on two publically available datasets with tuned hyper-parameters of MHAs methods for AD and MCI detection. • Enhancing AD detection performance by feature selection of present model and comparing research findings with previous techniques. [ABSTRACT FROM AUTHOR]

Smart cities have been widely investigated, and several algorithms, techniques, and protocols have recently been developed to serve smart city environments. Intelligent traffic management is one of the main research fields in this area. It aims to control and monitor the traffic over downtown and highway scenarios. The real-time traffic characteristics on road networks highly affect safety conditions and driving behaviors. Detecting hazardous traffic conditions and congested areas early helps drivers make the best decision to avoid dangerous situations. After extensive training, machine learning algorithms are mainly used for prediction in this field. In the intelligent traffic control target, trained machines are beneficial in obtaining instant and accurate predictions regarding traffic congestion levels and their distributions. Several parameters control the efficiency of the selected machines. This includes the dataset quality, the regression algorithm's parameters, and how they have been tuned. In this work, we first create high-quality datasets representing the moving vehicles for different main roads in various countries, including Bank St in Canada, North Rodeo Drive (NRD) in Los Angeles, USA, Queen Rania Al Abdullah (QRA) road in Jordan, and Sheikh Zayed Road (SZR) in Saudi Arabia. Then, we test the most popular regression algorithms and evaluate their performance in terms of temporal prediction of traffic characteristics and traffic congestion in different geographical road scenarios. We optimized these regression algorithms and tuned the used parameters using the grid search approach. After that, more advanced metaheuristic optimization algorithms were used to improve the prediction accuracy of the selected machine. Experimentally, we have proved the enhancements in predicting traffic characteristics for the tested road scenarios. Both approaches show enhanced results compared to using the regression algorithm without tuning the parameters. It shows advanced results for both approaches for the coefficient of determination measure, named R squared (R 2), with approximately 0.07, 0.04, and 0.08 enhanced values for Bank St., NRD, and SZR roads, respectively. In contrast, it shows approximately 0.95, 0.49, and 3.62 enhanced values for Bank St., NRD, and SZR roads, respectively, for the Root Mean Squared Error (R M S E) measure. [ABSTRACT FROM AUTHOR]

With the increase of mass customization, flexible job shop scheduling problem considering assembly stage has widely existed in many manufacturing industries, such as die-casting mould factories. This problem is to find a reasonable machine assignment and operation sequence both in fabrication and assembly stages and simultaneously maximize production efficiency. In reality, energy shortages and environmental pollution have given an impetus to the development of energy-aware production scheduling problems. In this study, we address an energy-aware flexible assembly job shop scheduling problem (EFAJSP) with the objectives of minimizing flow time and energy consumption and first develop a mixed-integer linear programming (MILP) model to solve EFAJSP problem. Then, the model-specific characteristics are extracted and applied to a matheuristic decoding method for exploring the Pareto optimal solution. Due to the complexity of EFAJSP problem, a matheuristic and learning-oriented multi-objective artificial bee colony algorithm (MLABC), which combines the advantages of mathematical programming, reinforcement learning and meta-heuristic algorithm, is proposed. In addition, an initialization, destruction/construction operator and population update operator are proposed and work together to improve the exploration and exploitation performance of the proposed MLABC. Finally, numerical experimental results demonstrate the effectiveness of the proposed MILP model and the superiority of the MLABC over other algorithms in the literature. [ABSTRACT FROM AUTHOR]

• The shear strength of corroded RC beams is predicted using ML approach; • Experimental data and Gradient Boosting (GB) as ML algorithm are utilized; • GB model is optimized with 3 hybrid algorithms, in which HGS is the best; • The optimized GB-HGS outperforms others and used to conduct sensitivity analysis; • Optimal beam solutions were provided as guidance for appropriate design; Shear strength (SS) is an essential component in the design of reinforced concrete structural elements, particularly in severe settings where reinforcements can occur and cause a loss in SS. The purpose of this work is to develop a framework for predicting the SS and optimal design of corroded reinforced concrete (CRCo) beams by using a machine learning (ML) approach. The model employed was Gradient Boosting (CGB), optimized using three metaheuristic algorithms. By optimizing the CGB model with the Hunger Games Search (HGS) algorithm, the study achieved the best ML model for predicting the SS of CRCo beams, with an R2 value of 0.996. The proposed model outperformed five other empirical SS models for CRCo beams, and sensitivity, partial dependence analyses were also conducted to explore the impact of various variables on SS. Finally, this work provided guidance on selecting appropriate beam sizes and corrosion rates for optimal CRCo beams design solutions. [ABSTRACT FROM AUTHOR]

• An intelligent framework is developed to frequently update set points in the grinding circuit, aiming to maximise SAG mill throughput. • The prediction model can be frequently retrained using data from process sensors and updated to adapt to changing conditions of the grinding process. This dynamic approach distinguishes it from static modelling methods, which may experience reduced accuracy over time. • CatBoost is the most accurate predictor of SAG mill throughput among 17 machine learning models evaluated. Differential Evolution outperforms other optimisation algorithms. In mineral processing plants, grinding is a crucial step, accounting for approximately 50% of the total mineral processing costs. Semi-autogenous grinding (SAG) mills are extensively employed in the grinding circuit of mineral processing plants. Maximising SAG mill throughput is of significant importance considering its profound financial outcomes. However, the optimum process parameter setting aimed at achieving maximum mill throughput remains an uninvestigated domain in prior research. This study introduces an intelligent framework leveraging expert knowledge, machine learning techniques and evolutionary algorithms to address this research need. In this study, an extensive industrial dataset comprising 36,743 records is utilised and relevant features are selected based on the insights of industry experts. Following the removal of erroneous data, an evaluation of 17 machine learning models is undertaken to identify the most accurate predictive model. To improve the performance of the model, feature selection and outlier detection are executed. The resultant optimal model, trained with refined features, serves as the objective function within three distinct evolutionary algorithms. These algorithms are employed to identify parameter configurations that maximise SAG mill throughput while adhering to the working limits of input parameters as constraints. Notably, analysis revealed that CatBoost, as an ensemble model, stands out as the most accurate predictor. Furthermore, differential evolution emerges as the preferred optimisation algorithm, exhibiting superior performance in both achieving the highest mill throughput predictions and ensuring robustness in predictions, surpassing alternative methods. [ABSTRACT FROM AUTHOR]

The study presents an innovative hybrid machine learning model tailored for predicting the load-displacement characteristics of bored piles, specifically those integral to high-rise buildings. Incorporating critical design parameters—diameter, length, Standard Penetration Test (SPT) indices, and effective overburden pressure—the model leverages a dataset of 1650 samples from static load tests in Vietnam. This hybrid approach integrates the Three Modified Symbiotic Organisms Search algorithm (3mSOS) with the Levenberg–Marquardt backpropagation neural network (LMNN) to establish the intricate relationship between these design parameters and the load-displacement response of the piles. Numerical results underscore the model's exceptional performance in accurately predicting the load-displacement behavior of bored piles. Rigorous validation employs an independent dataset derived from bidirectional pile load tests, affirming the model's reliability. A comprehensive sensitivity analysis provides valuable insights into the mechanisms governing load-bearing. Feature importance analysis and partial dependence plots reveal nuanced relationships among input variables and output behavior. The model's novelty lies in pioneering the application of advanced metaheuristic algorithms, notably 3mSOS, in pile foundations—a distinctive contribution to geotechnical engineering. This research holds significant promise for enhancing the efficiency and accuracy of pile design in high-rise buildings, thereby bolstering the overall reliability of foundation design. • 3mSOS-LMNN transforms large bored pile foundation design, ensuring precise load-displacement predictions. • SHAP analysis pinpoints parameters influencing pile head displacement, aiding informed decision-making. • PDPs provide a profound understanding of influential parameters, emphasizing their crucial role in pile behavior. [ABSTRACT FROM AUTHOR]

• Propose CycleGAN data architecture for securing the smart cities against FDIA. • Proposing IoT concept for improving real-time monitoring of smart cities. • Developing a new IDS hyperparameter adjusting system based on BBO and WOA. In the realm of securing smart cities against emerging cyber threats, this research encompasses three distinct yet interconnected initiatives. First, a pioneering data architecture design leveraging CycleGAN is proposed to counteract False Data Injection Attacks (FDIA). By cyclically transforming data distributions, CycleGAN ensures the integrity and reliability of smart city information, fortifying against malicious manipulations. Second, an innovative Internet of Things (IoT) concept is introduced, aiming to enhance real-time monitoring and context-aware threat detection within Intrusion Detection Systems (IDS). Harnessing the wealth of data generated by IoT devices, this concept provides a comprehensive understanding of network activities, fostering adaptive responses to emerging threats. Lastly, the research delves into the development of a novel IDS hyperparameter adjusting system. This system integrates the strengths of Biogeography-Based Optimization (BBO) and Whale Optimization Algorithm (WOA) to fine-tune IDS configuration parameters. Drawing inspiration from biogeographical principles and collaborative whale behavior, the hybrid optimization system balances exploration and exploitation, adapting the IDS to diverse network environments. Together, these initiatives represent a holistic approach to fortifying smart cities through cutting-edge data architecture, IoT-driven threat detection, and optimized IDS configurations, contributing to the resilience and cybersecurity of modern urban landscapes. [ABSTRACT FROM AUTHOR]

Graph meta-learning has recently received significantly increased attention by virtue of its potential to extract common and transferable knowledge from learning different tasks on a graph. Existing methods for graph meta-learning usually leverage local subgraphs to transfer subgraph-specific information. However, they inherently face the challenge of imbalanced subgraphs due to inconsistent node density and different label distributions over local subgraphs. This paper proposes an adaptive graph meta-learning framework (AG-Meta) for learning the consistent and transferable representation of a graph in a way that can adapt to imbalanced subgraphs. Specifically, AG-Meta first learns the structural representation of subgraphs with various degrees using an Adaptive Graph Cascade Diffusion Network (AGCDN). AG-Meta then employs a prototype-consistency classifier to produce more accurate transferable inductive representations (also called prototypes) under few-shot settings with different label distributions of a subgraph. In the context of optimizing a model-agnostic meta-learner, a novel metric loss is finally introduced to achieve structural representation and prototype consistency. Extensive experiments are conducted to compare AG-Meta against baselines on five real-world networks, which validates that AG-Meta outperforms the state-of-the-art approaches. • A novel adaptive graph meta-learning framework is proposed. • An ensemble weight scheme is used to deal with inconsistency in densities and labels. • A prototypical-consistency loss is developed for graph meta-learning. • Our AG-Meta model outperforms several baselines. [ABSTRACT FROM AUTHOR]

Genetic algorithms and machine learning methods can accurately anticipate hybrid nanofluids' complicated rheology. Scientists and engineers can understand hybrid materials by using genetic algorithms to optimize and machine learning to discover complicated relationships between input variables and rheological responses. As a continuation of the author's previous research on the rheological properties of a nano-lubricant based on engine oil and hybrid nanoparticles, this study uses machine learning and genetic algorithms to theoretically assess the dynamic viscosity of the MWCNT-MgO/oil SAE 50 hybrid nanofluid and identify optimal properties. MLR, D-Tree, Ridge, PLR, SVM, Lasso, ECR, GPR, and MPR are used for regression analysis. Best multi-objective issue solutions are represented by the Pareto front. The NSGA-II algorithm determines the Pareto front. The MPR and NSGA-II algorithms provide a Pareto front with the most precise optimal spot boundaries. The Weighted Sum Method (WSM) simplifies multi-objective problems into single-objective problems, making optimal solutions easier to find. The results show that the maximum margin of deviation for μ nf and τ is − 2.5615 and − 5.239, respectively. According to the Taylor chart, the best μ nf mode for R, RMSE and STD is equal to 0.9983, 7.6639, 130.0056. Also, these values for τ are equal to 0.9996, 15.4515, and 516.0219. • Rheological behavior of the MWCNT-MgO/SAE 50 hybrid nanofluid was studied. • use of meta-heuristic algorithm NSGA-Ⅱ to find the most optimal point that satisfies μ nf and τ. • Optimal solution is represented by a set of points known as the Pareto front. • The best algorithm in terms of the Taylor diagram for μ nf output is the MPR algorithm, and the worst is the Lasso algorithm. [ABSTRACT FROM AUTHOR]

This paper aims to introduce an artificial intelligence approach to enhance short-term daily streamflow forecasts through the integration of machine learning (ML) techniques with meta -heuristic (MH) algorithms. To maximize performance, a comprehensive analysis involving the autocorrelation and partial autocorrelation functions of the observed streamflow was conducted to derive ten distinct input combinations, from which the most suitable set was selected for investigation. Subsequently, six distinct ML methods, namely, decision trees, support vector regression, long short-term memory, gated recurrent unit, multi-layer perceptron, and radial basis function (as the optimal input combination), were leveraged to predict one-day-ahead streamflow. The most promising outcomes from these six explored methods were synergistically amalgamated to enhance predictive precision. This amalgamation was then optimized by utilizing parameter values hyper-tuned by four different MH algorithms. The algorithms included the shuffled frog leaping algorithm, particle swarm optimization, ant colony optimization (ACO), and gray wolf optimization (GWO). In culmination, a case study centered around the "Wick" river in the UK was conducted to rigorously evaluate the performance of the proposed method, utilizing a range of statistical evaluation indices. The results underscored the efficacy of the proposed combined method in substantially refining streamflow predictions, exhibiting an impressive level of accuracy. Additionally, the outcomes emphasized that ACO and GWO stand out as the optimal algorithms for calibrating the parameters of the amalgamated model. For instance, the R2 values attained by the combined models optimized through the ACO and GWO algorithms were 0.69422 and 0.69444, respectively. These values notably surpass those observed in the initial models, substantiating the superiority of the applied optimization approaches. [ABSTRACT FROM AUTHOR]

As wireless communication technology continues to advance, the antenna, as an essential front-end device in radio communication system, is surrounded by more and more complex electromagnetic wave environments with increasing variety, resulting in greater demand for antennas and higher design requirements. While the traditional antenna design methods suffered from the disadvantage of low design efficiency, a powerful tool for accelerating antenna design is the modelling of antennas with neural networks. Aiming to enhance the modeling accuracy of neural network, multiple novel meta-heuristic swarm intelligent algorithms are introduced and part of them are modified for the purpose of applying to optimizing network's weights and biases so as to raise the antenna model's prediction precision on the basis of neural network. Specifically, the intelligent algorithms and their improvement directions include the strategy of optimizing weights and biases for neural networks with seagull optimization algorithm, optimizing the weights and biases of neural network with the improved butterfly algorithm fused with reverse learning, and the artificial rabbit algorithm optimizing the neural network weights and biases. In addition, two intelligent optimization algorithms that are already more mature: the particle swarm algorithm and the genetic algorithm are added to compare with the above three algorithms. The accuracy of neural network prediction before and after the optimisation of neural network by seagull algorithm, the butterfly algorithm incorporating reverse learning, the artificial rabbit algorithm, the particle swarm algorithm, and the genetic algorithm are got through the results respectively. The results of the experiments displayed that the neural network optimized of the improved butterfly algorithm incorporating reverse learning has a prediction accuracy of 99.69 % with stable results, the optimised neural network prediction accuracy of the seagull algorithm reaches 99.51%, and the optimised neural network prediction accuracy of the artificial rabbit algorithm is 99.49 %. The remaining two traditional algorithms optimized neural network accuracy is 83.1 % and 99.43 % respectively. Therefore, the improved butterfly algorithm incorporating reverse learning is the most effective among these three new algorithms applied to the field of antenna prediction. Moreover, the running time of the network optimized by different algorithms is quite distinct, among which the neural network optimized by the improved butterfly algorithm incorporating reverse learning takes the shortest time, which increases the prediction efficiency of the network by more than 70%. In summary, the application of the fused reverse learning improved butterfly algorithm in optimizing neural network predictions yields the shortest processing time and highest accuracy. This not only enables faster and more precise antenna design but also holds greater significance for the field of antenna design and analysis. • Improved butterfly algorithm combined with BPNN. • Integration of reverse learning with the butterfly algorithm for optimizing neural networks. • A large amount of antenna data extracted in HFSS is used to train the neural network constructed after optimization with an intelligent algorithm. • The new network achieves 99.57% prediction efficiency. [ABSTRACT FROM AUTHOR]

Hyperbolic spaces have recently achieved acceleration in the context of machine learning of their high capacity and tree-likeliness structures, taxonomies, text, and graphs. With the same dimension, a hyperbolic vector can represent richer information than a Euclidean vector. In this paper, we propose Metaheuristic optimization algorithms in hyperbolic space. Considering that the most popular optimization tools have not been generalized in hyperbolic space, we design optimization algorithms according to the specific property of the hyperbolic manifold. However, a major bottleneck here is the obscurity of hyperbolic space and a better comprehension of its gyrovector operations. We aim to introduce researchers and practitioners in the metaheuristic community to the hyperbolic equivariant of the Euclidean operations necessary to tackle their application to Metaheuristic optimization. We conduct experiments on various metaheuristic algorithms in hyperbolic space. [ABSTRACT FROM AUTHOR]

Groundwater (GW) salinization of coastal aquifers has become a serious problem for attaining sustainable water resource management in Saudi Arabia and other parts of the world. Therefore, it is crucial to assess the extent of this salinization to protect and manage our water resources effectively. This research proposed real fieldwork GW samples at several locations supported with experimental based on chromatography (IC) and inductively coupled plasma mass spectrometry (ICP-MS) to analyze several GW physical, chemical, and hydro-geochemical elements. In this study, we model GW salinization with machine learning algorithms such as support vector regression, gaussian process regression, artificial neural networks, and least squares ensemble boosting regression tree. The performance of the standalone models was optimized with metaheuristic optimization-based algorithms such as fuzzy hybridized genetic algorithm (ANFIS-GA) and particle swarm optimization (ANFIS-PSO). The outcomes based on three variable input combinations were validated using several performance indicators and graphical methods. The quantitative analysis indicated that GPR-Combo1(MAE = 0.006 mg/L), Ensm- Combo2 (MAE = 0.025 mg/L), and GPR- Combo3 (MAE = 0.078 mg/L) proved merit among the standalone combinations. Where combo 1, 2, and 3 stand for model combinations derived from feature selection. The cumulative probability function (CPF) demonstrated that heuristic optimization ANFIS-GA (MAE = 0.0025 mg/L, MAPE = 0.19183) and ANFIS-PSO (MAE = 0.0018 mg/L, MAPE = 0.0723) outperformed the standalone error accuracy and served reliable approach. Both the standalone models and heuristic algorithms used for GW salinization modeling have demonstrated promising results in accurately predicting salinity. This approach could aid in effectively managing the GW resources for sustainable development. • GW salinization addressed by field, experiment, and Al-based modelling. • Standalone hybrid techniques and heuristic algorithms were developed. • The nature-inspired optimization techniques proved reliable. • Na, Cl, Br, and Ca are the most important parameters for GW salinization modelling. • Reliable results and possible research directions were summarized for decision-makers. [ABSTRACT FROM AUTHOR]

Droplet-based microfluidics holds promise for advanced biomedical applications due to its enhanced accuracy, throughput, rapid reaction time, and minimal reagent consumption. However, achieving droplets with the desired radius often entails costly design iterations. In this paper, we propose a novel approach that combines machine learning algorithms with metaheuristic algorithms to leverage their respective strengths in speed and accuracy. Specifically, we demonstrate the effectiveness of the Rain Optimization Algorithm (ROA) in enhancing the accuracy of five machine learning models. Additionally, we introduce a hybrid model, optimized by ROA, for predicting droplet generation in a cross-junction microfluidic channel, achieving a coefficient of determination (R2) of 0.937, reflecting a 3.6 % increase in accuracy. We employ symbolic regression to validate the hybrid model, yielding an R2 value of 0.987, indicating strong agreement with real data. Furthermore, we utilize the SHAP method in Explainable AI to interpret the hybrid model's performance. Notably, the proposed hybrid model requires only four seconds to solve the prediction problem, significantly reducing computational time compared to the numerical model and experimental method. This approach facilitates the study of each input parameter's impact on droplet size prediction, offering otherwise challenging and costly insights to obtain experimentally. [ABSTRACT FROM AUTHOR]

The visual abstract for the corn leaf disease prediction is delineated. [Display omitted] • Proposing JSWOA-optimized hybrid 3DCNN-RNN to predict various corn leaf diseases. • Integrating 3DCNN, RNN and LSTM models to classify 8 corn leaf diseases. • Hyperparameters of hybrid 3DCNN-RNN model are optimized by JSWOA. • JSWOA increases the global search ability and convergence speed. • The hybrid model reaches above 90% performance in predicting corn leaf disease. More than 30 % of corn is produced by the United States, which is the world's leading corn producer. The deficiency in the corn leaf may influence crop yield manufacturing. Moreover, those methods only serve yield assessment and do not compute the prediction uncertainties in the crop. To overcome such limitations, the hybrid model called Whale Optimization Algorithm with Joint Search Mechanisms (JSWOA) optimized hybrid 3DCNN-RNN model is proposed for the corn leaf disease prediction. 3DCNN includes Four MaxPooling3D layers and six 3D convolutional layers to enhance the identification of 3-Dimensional images, and it determines the temporally local features and spatial features. JSWOA optimized hyperparameters of the hybrid 3DCNN-RNN models. For performing different experiments based on machine learning, Maize_in_field, and KaraAgro AI maize are the two datasets employed. The performance rate of MCC, recall, precision, accuracy, specificity, F1-score, and AUC are assessed to ensure the efficacy of the proposed hybrid model. Also, the existing CNN, optimized DenseNet, and AlexNet models are analyzed for performance comparison. The simulation outcome reveals that the proposed hybrid model reaches above 90 % performance in predicting various maize leaf classes on both datasets. Therefore, the state-of-the-art method is outperformed by the proposed hybrid model. [ABSTRACT FROM AUTHOR]

This research presents a comprehensive, sequential ensemble framework meticulously crafted for optimizing investment portfolios, focusing on the construction industry. It employs decision tree-based and metaheuristic optimization algorithms to create an efficient stock-selection framework grounded in financial analysis. This approach offers a profitable investment strategy integrated with portfolio optimization while systematically lowering the portfolio investment threshold. In this framework, a web crawler was deployed to gather daily closing prices of stocks from the Taiwan Stock Exchange, along with monthly revenue data and financial statements. Subsequently, decision tree-based models were utilized to pinpoint the fundamental financial indicators with significant explanatory power over revenues. The stock selection conditions aligned with these indicators were optimized through a newly developed metaheuristic algorithm named the forensic-based investigation (FBI) optimizer. The optimal conditions were subsequently integrated with the equal-weighting scheme, mean-variance method, and hierarchical risk parity to identify the most effective investment portfolio strategies. Backtesting results showed that the proposed stock-portfolio investment strategy, optimized through machine learning and a metaheuristic algorithm, is well-suited for construction and all-stock categories. This study equips professional investment advisors or securities investment institutions with a decision-aid expert system for initial stock selection. • Developing an innovative stock selection strategy based on specific criteria thresholds. • Utilizing machine learning models (RF, XGBoost, and LightGBM) to analyze key financial indicators. • Merging financial analysis with a Forensic-Based Investigation (FBI) algorithm for portfolio construction. • Examining portfolio allocation strategies (EW, MV, and HRP) to maximize investment returns. • Showcasing excess returns through strategic portfolio management in backtesting investments. [ABSTRACT FROM AUTHOR]

High hydrogen production is essential in sorption-enhanced steam methane reforming (SE-SMR). This study presents an innovative optimization framework to optimize the H 2 yield of SE-SMR through a coupled reaction model, prediction surrogate model, and whale optimization algorithm. Firstly, a computational fluid dynamics chemical reaction model of SE-SMR is constructed. A database is established to consider the H 2 yield of SE-SMR with temperature, pressure, velocity, and steam to carbon ratio. Then, a surrogate model is developed on the basis of eXtreme Gradient Boosting (XGBoost) to predict the H 2 yield under various operating parameters. Finally, the surrogate model is exploited to calculate the fitness function of the whale optimization algorithm to construct an optimization framework for maximizing the H 2 yield. The results show that the XGBoost ensemble surrogate model can predict the H 2 yield of SE-SMR within 4 s with high accuracy. The optimization framework can optimize the maximum H 2 yield and the corresponding operating parameters within 1 min. The optimization results are validated by applying the model and the percentage error of only 0.79 %. Moreover, under optimal operating conditions, methane is almost completely converted to hydrogen, and the dry hydrogen mole fraction could reach 95.96 %. The presented optimization framework can guide the prediction of H 2 yield and the optimization of operating parameters in SE-SMR. [Display omitted] • A novel framework coupling surrogate model and WOA algorithm is proposed. • A database is constructed by developing SE-SMR chemical reaction model. • H 2 yield prediction surrogate model is designed using XGBoost algorithm. • Optimal operating condition is obtained through the proposed novel framework. [ABSTRACT FROM AUTHOR]

Traditional machine learning algorithms face challenges in assessing flood susceptibility reliably due to their low robustness and the inherent 'black-box' nature. This paper utilizes five hyperparameter optimization algoirthms (HPO), namely grid search (GS), random search (RS), gauss process (GP), tree-structured parzen estimator (TPE) and simulated annealing (SA), to tune the traditional random forest's (RF) hyperparameters to improve the robustness of flood hazard mapping (FHM) models at Ningxiang City Hunan Province, China. Additionally, SHapley Additive exPlanations (SHAP) method were used to interpret the decision-mechanisms of these flood hazard models. This study considers 19 pluvial flood influencing factors and 2064 flood locations to create a geospatial database. The performance of each hybrid model was evaluated by area under the receiver operating characteristic (ROC) curve (AUC) and several validation methods. The results demonstrate that the developed hybrid models demonstrated good performance, with RF-TPE achieving the highest AUC (0.9660), followed by RF-GP (0.9648), RF-SA (0.9624), RF-GS (0.9612), RF-RS (0.9600), and RF (0.9539). The RF-TPE model exhibits superior robustness than other models, and the FHM constructed using it is more reliable. HPO is an effective approach to improve the predictive accuracy and robustness of FHM models. When considering limited computational resources, Bayesian optimization (TPE) should be prioritized for optimizing FHM models, followed by metaheuristic algorithms and model-free algorithms. Moreover, the study revealed that distance from river, peak rainfall intensity, continuous rainfall, antecedent effective rainfall, and terrain relief, are the most significant for pluvial FHM modeling in this region. [ABSTRACT FROM AUTHOR]

This paper proposes an innovative method for achieving optimal tracking control in nonlinear continuous-time systems with input constraints. The method combines reinforcement learning and hybrid metaheuristics to enhance the controller's performance. Specifically, a hybrid metaheuristic algorithm is employed to optimize the hyperparameters of a critic neural network, which serves as the system's controller. The proposed approach is evaluated through extensive simulation studies on a nonlinear system with input constraints. Results demonstrate its superiority over traditional control techniques in terms of accuracy, timeliness, and stability. Notably, the approach effectively eliminates overshoot and steady-state error while providing precise and prompt tracking and showcasing remarkable robustness against model uncertainties. By synergistically integrating reinforcement learning and hybrid metaheuristics, this approach represents a significant advancement in enhancing the control performance of complex nonlinear systems. The simulation studies confirm superiority of the proposed approach over existing techniques, offering a promising solution for achieving optimal tracking control in nonlinear systems with input constraints. This approach holds potential for a wide range of applications, including robotics, aerospace, and manufacturing, where precise and prompt tracking control is critical. [ABSTRACT FROM AUTHOR]

This study presents an artificial intelligence-based model to predict the flexural strengthening of RC beam/slab strengthened with UHPC (RC-UHPC) and perform uncertainty quantification of such values. Improved eliminate particle swamp optimization hybridized artificial neural network (IEPANN) was a model developed on artificial neural network (ANN) optimized by improved-eliminate-particle-swamp optimization (IEPSO). Experimental datasets were used to establish the hybrid ML model. The efficiency of IEPANN was compared against support vector regression (SVR) and multiple linear regression (MLR). The IEPANN demonstrates outstanding performance over MLR and SVR. The flexural strengthening of RC-UHPC depends on materials and geometrical configurations, which have stochastic characteristics. Convergence studies, sensitivity, and probabilistic analyses were conducted on each random parameter. Simulation results indicated that strengthening configuration is the most sensitive to flexural strengthening. Systematic assessment of uncertainties in flexural strengthening may lead to a higher confidence level in model outcomes, and RC-UHPC is examined and designed. It enhances construction reliability. [ABSTRACT FROM AUTHOR]

Flue gas water-alternating gas (flue gas-WAG) is a promising technique for enhancing oil production and reducing greenhouse gas emissions. The effective utilization of this technique relies on identifying the optimal factors, often determined through numerous numerical simulations. This paper introduces a cost-effective optimization framework that integrates machine learning models and diverse optimization algorithms to identify the best parameters for injecting flue gas-WAG into the "Gullfaks" reservoir. Robust Machine learning models, including Multilayer Perceptron, Cascade Forward Neural Network (CFNN), Radial Basis Function, and Generalized Regression Neural Network, were employed as proxy models. The CFNN model, with satisfactory agreement with actual data (average absolute relative error of 0.4543 % for oil recovery factor (RF) and 1.9366 % for CO 2 storage), is selected for optimization through metaheuristic algorithms, including Non-dominated sorting genetic algorithm version II (NSGA-II), Pareto envelope-based selection Algorithm version-II (PESA_II), Multi-objective particle swarm optimization (MOPSO), and Multi-objective gray wolf optimization (MOGWO(. Among the optimization algorithms, MOGWO outperforms others in terms of speed and accuracy, yielding Pareto-optimal solutions with an RF of 0.8284 and a CO 2 storage of 1.5314 × 107 (kg-mol). The proposed approach uses Pareto dominance for insightful field development planning, enabling decision-makers to choose flue gas-WAG parameters based on future circumstances. [Display omitted] • A robust optimization framework is proposed for flue gas storage and enhanced oil recovery. • NSGA-II, MOPSO, MOGWO and PESA-II algorithms are utilized to optimize the water-alternating flue gas process. • CFNN, MLP, RBFNN, and GRNN are developed as proxy models that imitate the numerical simulator. • CFNN outperforms all developed models with AAPRE of 0.4543 % and 1.9633 % to estimate oil recovery factor and CO 2 storage, respectively. • The created Pareto fronts allow flexible applicability conditions to examine the design factors of the process. [ABSTRACT FROM AUTHOR]

[Display omitted] • Detection by terahertz time-domain spectroscopy combined with chemometrics. • Whale Optimization Algorithm was used to improve RF and ELM model. • Different concentrations of PDE5 Inhibitor can be detected. Combating the illicit use of PDE5 inhibitor drugs is a focal point in forensic science research. In order to achieve rapid identification of such drugs, this study applies terahertz time-domain spectroscopy combined with chemometrics to establish a fast and accurate detection method for PDE5 inhibitors. The optimal detection method is determined by comparing the spectral performance of three optical parameters, namely absorption coefficient, refractive index, and dielectric constant. Linear discriminant models based on different spectral parameters, whale optimization algorithm optimized extreme learning machine models, and whale optimization algorithm optimized random forest models are established. The effectiveness and performance of principal component analysis and competitive adaptive reweighted sampling algorithm for spectral feature data selection are also investigated. The PDE5 inhibitor identification model based on the competitive adaptive reweighted sampling - whale optimization algorithm - random forest (CARS-WOA-RF) model achieves an accuracy of 98.61%, and the identification model for two concentrations of Sildenafil achieves 100% accuracy. The results demonstrate that terahertz time-domain spectroscopy combined with chemometrics can effectively detect various common types of PDE5 inhibitor drugs and different concentrations. [ABSTRACT FROM AUTHOR]

This comprehensive review focuses on the Autonomous Driving System (ADS), which aims to reduce human errors that are the reason for about 95% of car accidents. The ADS consists of six stages: sensors, perception, localization, assessment, path planning, and control. We explain the main state-of-the-art techniques used in each stage, analyzing 275 papers, with 162 specifically on path planning due to its complexity, NP-hard optimization nature, and pivotal role in ADS. This paper categorizes path planning techniques into three primary groups: traditional (graph-based, sampling-based, gradient-based, optimization-based, interpolation curve algorithms), machine and deep learning, and meta-heuristic optimization, detailing their advantages and drawbacks. Findings show that meta-heuristic optimization methods, representing 23% of our study, are preferred for being general problem solvers capable of handling complex problems. In addition, they have faster convergence and reduced risk of local minima. Machine and deep learning techniques, accounting for 25%, are favored for their learning capabilities and fast responses to known scenarios. The trend towards hybrid algorithms (27%) combines various methods, merging each algorithm's benefits and overcoming the other's drawbacks. Moreover, adaptive parameter tuning is crucial to enhance efficiency, applicability, and balancing the search capability. This review sheds light on the future of path planning in autonomous driving systems, helping to tackle current challenges and unlock the full capabilities of autonomous vehicles. [ABSTRACT FROM AUTHOR]

• A feature selection method based on binary dandelion algorithm is proposed. • The algorithm applies a sine-cosine operator and a restart strategy. • Mutual information and quick bit mutation improve the performance of the algorithm. • The algorithm performs well on datasets with larger dimensions. Feature selection (FS) is an important data preprocessing technology for machine learning and data mining. Metaheuristic algorithm (MH) has been widely used in feature selection because of its powerful search function. This paper presents an improved Binary Dandelion Algorithm using Sine Cosine operator and Restart strategy (SCRBDA) for feature selection. First, the sine cosine operator is used in the radius formula of the core dandelions (CD), which significantly enhances the ability of algorithm development and exploration. Secondly, the algorithm uses a restart strategy to increase its ability to get rid of local optimum. Thirdly, mutual information is used to guide the generation of some dandelions, which pays more attention to the correlation between the selected features and categories. Finally, quick bit mutation is used as the mutation strategy to improve the diversity of the population. The SCRBDA proposed in this paper was tested on 18 datasets of different sizes from UCI machine learning database. The SCRBDA was compared with 8 other classical feature selection algorithms, and the performance of the proposed algorithm was evaluated through feature subset size, classification accuracy, fitness value, and F1-score. The experimental results show that SCRBDA achieves the best performance, which has stronger feature reduction ability and achieves better overall performance on most datasets. Especially on large-scale datasets, SCRBDA can obtain extremely smaller feature subsets while maintaining much higher classification accuracy, and satisfactory F1-score. [ABSTRACT FROM AUTHOR]

• Data-driven metamodel developed for ZnO varistor sintering. • ML algorithms used for non-normal distribution of permittivity values. • HMA used for process parameter optimization. • Integration of Hybrid-EDT into HMA saves time and resources. • Successful validation experiment conducted under optimal conditions. A ZnO varistor functions as a circuit protection device against surge voltage because of its nonlinear current/voltage characteristics. Achieving the desired electrical properties in a ZnO varistor necessitates meticulous control over its microstructure, which is achieved by regulating parameters during the sintering process, such as the sintering temperature, sintering time, heating rate, and cooling rate. In this study, a metamodel was developed through machine learning using a dataset obtained from a laboratory-scale furnace by employing the design of experiment approach and incorporating the aforementioned four process variables and permittivity. A hybrid metaheuristic optimization algorithm was then applied to determine the optimal process conditions, maximizing permittivity through the metamodel. To adapt the derived optimal conditions to a scale-up sintering furnace, temperature data were collected from laboratory-scale and scale-up sintering furnaces through numerical simulation. The permittivity of the sintered ZnO varistor, which was predicted by inputting the corrected process variables into the metamodel, exhibited a 4.6% difference from the experimental value, representing the average permittivity of ZnO varistors sintered in the scale-up furnace. The application of this simulation-based virtual scale-up to ceramic processing can significantly reduce the development time required for scale-up processes. [ABSTRACT FROM AUTHOR]

The proactive prediction and systematic classification of students' academic performance empower educational administrators with the invaluable capability to not only pinpoint potential challenges but also to craft targeted strategies and interventions that are tailored to the unique needs of students. This multifaceted approach enables educational institutions to proactively address issues within the education system, fostering a more equitable and effective learning environment for all, while simultaneously fostering a culture of continuous improvement and accountability in the pursuit of educational excellence. Hence, the current investigation aims to classify and predict the students' performance by examining and comparing the machine learning and artificial neural network assessments. Five methods of the Random Forest Classifier, the Decision Tree Classifier, the K Neighbors Classifier, the MLP Classifier, and the XG-Boost Classifier are used. These methods' performances are compared through the accuracy, precision, recall, and F1-score indicators. This comparison is applied to the base data and balanced data, which is carried out by the SVM-SMOTE technique. Finally, five metaheuristic algorithms are applied to the selected method to evaluate the performance indicators of the hybrid models. The results indicate that applying the SVM-SMOTE technique improves the methods' performance, in which the XG-Boost represented the best performance. As a result, the metaheuristic algorithms are applied to the XG-Boost, yielding to 9.33%, 8.44%, 9.33%, and 9.27% enhancement of the Accuracy, Precision, Recall and F1-Score values. Subsequently, the Enhanced Artificial Ecosystem-Based Optimization + XG-Boost hybrid method provides the accuracy and F1-score values of 0.9417 and 0.9413. These results underscore the potential of combining machine learning techniques with metaheuristic algorithms to enhance the accuracy and effectiveness of predicting and classifying students' performance, thus providing valuable insights for educational administrators to address issues and improve the education system. [ABSTRACT FROM AUTHOR]

[Display omitted] Feature selection (FS) is a crucial process that aims to remove unnecessary features from datasets. It plays a role in data mining and machine learning (ML) by reducing the risk associated with high-dimensional datasets. FS is considered a challenging problem that is difficult to solve efficiently due to its combinatorial nature. As the size of the problem increases, the computation time also grows. Recently, researchers have focused on metaheuristic FS algorithms specifically designed for high-dimensional datasets. Therefore, this article proposes a powerful metaheuristic algorithm called Binary Enhanced Golden Jackal Optimization (BEGJO), which is an improved version of the recently published Golden Jackal Optimization (GJO) algorithm. The original GJO algorithm faces challenges when dealing with high-dimensional FS problems, as it tends to get trapped in local optima. To address this issue, various enhancement strategies are employed to improve the efficiency of GJO. The proposed BEGJO algorithm utilizes Copula Entropy (CE) to reduce the dimensionality of high-dimensional FS problems while maintaining high classification accuracy using the K-Nearest Neighbour (K-NN) classifier. Additionally, four enhancement strategies are incorporated to enhance the exploration and exploitation capabilities of the fundamental GJO algorithm. The BEGJO algorithm is transformed into its binary form using the sigmoid transfer function, aligning it with the nature of the FS problem. It is then tested on various high-dimensional benchmark datasets. The effectiveness of BEGJO is evaluated by comparing it with well-known algorithms in terms of classification accuracy, feature dimension, and processing time. BEGJO outperforms other algorithms in terms of classification accuracy and feature dimension and ranks up to fourth in terms of processing time. Furthermore, the advantageous use of CE is demonstrated by comparing the performance of the proposed algorithm with traditional FS algorithms. Statistical evaluations are conducted to further validate the effectiveness and superiority of the proposed algorithm. The results confirm that BEGJO is an effective solution for high-dimensional FS problems. [ABSTRACT FROM AUTHOR]

Accurate prediction of heating and cooling loads in residential buildings is crucial for both researchers and practitioners. This study employs advanced forecasting techniques, utilizing support vector regression and extreme gradient boosting methods, enhanced by state-of-the-art metaheuristic algorithms. The metaheuristic optimizers applied include the Satin Bowerbird Optimizer, Ant Lion Optimizer, Artificial Ecosystem-Based Optimizer, Slime Mold Optimizer, Moth-Flame Optimizer, and Particle Swarm Optimizer. These optimizers are strategically used to refine the forecasting models, optimizing their parameters for increased precision. To evaluate the models, a suite of statistical indices is used, including mean square error, root mean square error, Mean Absolute Percentage Error, Mean Absolute Error, Relative Absolute Error, coefficient of correlation, Coefficient of Determination, and Normalized Mean Square Error. This comprehensive analysis assesses the effectiveness of the forecasting methods. The study's findings highlight that the combination of the Satin Bowerbird Optimizer and extreme gradient boosting is particularly effective, achieving high coefficients of determination for cooling and heating load predictions (0.95826 and 0.938048, respectively). This hybrid algorithm shows remarkable performance, with minimal error values and consistent convergence across various datasets. In summary, this research underscores the efficiency of integrating sophisticated machine learning models with metaheuristic optimization techniques, particularly identifying a potent hybrid algorithm for the accurate prediction of heating and cooling loads in residential buildings. • Proposing hybrid methods for the buildings heating and cooling loads prediction. • A combination of the SVR and XGBoost and metaheuristic algorithm has been proposed. • Six methaheuristic algorithm's accuracy has been compared via different statistical indexes. • The XGBoost combination is the most accurate method in term of the cooling and heating loads forecasting. [ABSTRACT FROM AUTHOR]

Different media filters and filtration media are used to eliminate suspended particles in microirrigation and therefore prevent clogging in the emitter. The water volume by filtration cycle is a parameter related to the filter and media capacity to retain particles, while turbidity is a variable related to particles in suspension in the water. Since turbidity can be measured easily and quickly, it is commonly mentioned in recommendations for the reuse of effluents in microirrigation. There are currently no models that are reliable enough to predict the filtered volume in each irrigation filter and outlet turbidity when using different filter types and media configurations. The object of this work was to propose a model that can detect early the filtered volume and the turbidity at the outlet values. This investigation presents an effective machine learning method, the Random Forest regression (RFR) in combination with the population-inspired metaheuristic optimization algorithm, called Differential Evolution (DE), for estimating the output turbidity and the filtered volume from a dataset with 1,016 samples of distinct media filters that use reclaimed effluent. The same experimental dataset was also fitted with Elastic-net, Lasso and Ridge regression machine learning methods also in combination with DE optimizer for comparison. This optimization performs the parameter tuning in the RFR using the training dataset, which considerably improves the accuracy of the regression. To achieve this, the most relevant operation input variables are tracked and analyzed: the kind of medium and filter, filtration velocity (v), height of the filter bed (H), cycle duration and the electrical conductivity for the filter inlet (EC i), pH i , dissolved oxygen (DO i), water temperature (T i) and the input turbidity (Turb i). There are two kinds of results. Firstly, the importance ranking of the input variables on the outlet turbidity and filtered volume is presented using the DE/RFR model. Secondly, an innovative model for the prediction of the outlet turbidity and filtered volume was built and a regression with optimized parameters was done and coefficients of determination of 0.9331 and 0.8712 for filtered volume and outlet turbidity were obtained with this DE/RFR–based model, respectively. Additionally, the outcomes from the Elastic-net, Lasso and Ridge models are worse than DE/RFR–relied model estimations. The DE/RFR-based model's strong performance was confirmed by the agreement between experimental data and the latter results. • The DE/RFR - based model is built as a predictive model of the outlet turbidity and filtered volume in granular filters. • The relative importance of the input operation variables in this process are determined. • The RFR results for the outlet turbidity and filtered volume are compared with the Elastic-net, Lasso and Ridge regressions. • The obtained correlation coefficients from this hybrid RFR–based model are about 0.97 and 0.94, respectively. [ABSTRACT FROM AUTHOR]

This study proposes different standalone models viz: Elman neural network (ENN), Boosted Tree algorithm (BTA), and f relevance vector machine (RVM) for modeling arsenic (As (mg/kg)) and zinc (Zn (mg/kg)) in marine sediments owing to anthropogenic activities. A heuristic algorithm based on the potential of RVM and a flower pollination algorithm (RVM-FPA) was developed to improve the prediction performance. Several evaluation indicators and graphical methods coupled with visualized cumulative probability function (CDF) were used to evaluate the accuracy of the models. Akaike (AIC) and Schwarz (SCI) information criteria based on Dickey-Fuller (ADF) and Philip Perron (PP) tests were introduced to check the reliability and stationarity of the data. The prediction performance in the verification phase indicated that RVM-M2 (PBAIS = -o.0465, MAE = 0.0335) and ENN-M2 (PBAIS = 0.0043, MAE = 0.0322) emerged as the best model for As (mg/kg) and Zn (mg/kg), respectively. In contrast with the standalone approaches, the simulated hybrid RVM-FPA proved merit and the most reliable, with a 5 % and 18 % predictive increase for As (mg/kg) and Zn (mg/kg), respectively. The study's findings validated the potential for estimating complex HMs through intelligent data-driven models and heuristic optimization. The study also generated valuable insights that can inform the decision-makers and stockholders for environmental management strategies. [Display omitted] • New hybrid machine learning (ML) models were developed for heavy metals prediction. • As (mg/kg) and zinc Zn (mg/kg) in marine sediments Algeciras Bay were predicted. • Akaike and Schwarz information criteria were used for data stationarity appraisal. • The proposed hybrid ML model exhibited the superior prediction performance. • The study provides insights for environmental decision-makers for proper management. [ABSTRACT FROM AUTHOR]