Search

Because of its high-speed flight and wide field of view, UAV can carry out comprehensive monitoring and target searching in a wide area, high altitude and long distance environment, and become an important tool of modern information inspection. Drone inspection technology has been widely used in many industries such as power, logistics, agriculture and security, especially in forest inspection, which plays a key role in protecting forest resources and maintaining ecological balance. The traditional manual inspection method has some problems, such as low efficiency, high missing rate and personnel safety risk, so it is difficult to meet the needs of modern inspection. This paper presents and implements a UAV inspection system based on computer vision. The system carries out inspection through the autonomous route planning of the UAV, collects image data and transmits it to the embedded device for analysis to extract the target monitoring information. Finally, the system generates the corresponding work order and sends it to the client, realizing the efficient, accurate and safe UAV inspection. This paper not only optimizes the UAV inspection algorithm design, but also improves the accuracy and efficiency of image recognition by applying the maximum likelihood estimation method, which provides reliable technical support for various inspection tasks. [ABSTRACT FROM AUTHOR]

• Single screw water vapor compression system was proposed and constructed, which can operate in a wide range of suction temperatures and compression ratio. • Variation of both the volumetric and isentropic efficiency with the independent variable factor were experimentally study. • Energy efficiency analysis was adopted to look for the main factors that affected the efficiency of single screw water vapor compressor. • Mathematical models of the single screw water vapor compressor were developed and optimized, to predict thermodynamic performance and optimize process. Single screw water vapor compressor has greater advantages under conditions of low suction temperature and high compression ratio, however, the current study has the following limitations: (1) the volumetric and isentropic efficiency were substituted as a fixed value into the models of compressor, (2) The suction temperature is concentrated in the medium-high region, (3) The compression ratio is concentrated in the medium-low range, which failed to accurately describe the water vapor compression process. Therefore, the relationship between the volumetric efficiency, isentropic efficiency and key characteristic parameters of compressor is studied experimentally, the mathematical models of the volume flow and compression power are established and verified by experiments, the changes of the mass flow, compression power, COP and SMER with suction temperature, discharge pressure and compression ratio are analyzed, the operation process parameters of the single-screw water vapor compressor are optimized. [ABSTRACT FROM AUTHOR]

With the development of the petrochemical industry, corrosion problems of special equipment occur frequently. How to achieve accurate prevention and control of equipment flow corrosion problems become a key factor in ensuring the safety of equipment processes. This paper combines industrial field operation data, through the isolated forest algorithm for the screening of abnormal values, focusing on the analysis of the type of corrosion in the heat exchanger. A Particle Swarm Optimization optimized Radial Basis Function (PSO-RBF) prediction model was constructed to achieve fast prediction of ammonium crystallization temperature, and the Particle Swarm Optimization (PSO) algorithm was optimized using an adaptive weight optimization scheme. It shows that the optimized model error is within 3%, while the prediction accuracy is improved by 9.54%, and the coefficient of determination has also been improved. This paper also compares and analyzes several mature algorithms commonly used today, and the results show that the model has significant advantages in terms of prediction performance. In addition, the model can be applied to the condition inspection of petrochemical equipment, providing a practical guarantee for equipment operation, and reducing the risk of enterprise operation and maintenance. [ABSTRACT FROM AUTHOR]

Experimental data under a wide range of conditions are essential for the optimization of combustion kinetic models. However, some laboratory measurements under given conditions may not be conducted due to the constraint of existing techniques. It is thus needed to employ the experimental data obtained under alternate conditions to improve the model predictions for the desired conditions. In this work, an active subspace-based similarity analysis is proposed as an experimental design method to find substitutes for experiments (or measurements) that are difficult to conduct. The eigenvalues and eigenvectors of the matrix that contains the gradient information of a model output with respect to inputs (matrix C of the active subspace) are used to calculate the cosine-based similarity of key reactions of two model targets. The method is demonstrated in three combustion systems, i.e. , ignition of hydrogen/oxygen mixture, premixed flame of the dimethyl ether (DME), and C 2 H 6 /O 2 systems in different reactors. The results show that if the similarity coefficient is large, the key reactions for the two model targets are similar, and the measurement of one target can improve the model prediction of the other target. In addition to designing experimental targets or conditions with strong constraint effects beforehand, this method can also be used to classify potential experimental targets/conditions. [ABSTRACT FROM AUTHOR]

This study is targeted to use AI (artificial intelligence) to predict the English proficiency test scores (CET-4) of students in private universities. The study will train and test multiple machine learning models using a dataset of students' demographic information, English language background, and test scores. These models will use different algorithms such as linear regression, decision trees, and neural networks to identify patterns and relationships in the data and make accurate predictions. The study aims to develop a reliable and effective predictive tool for educators to assess students' language proficiency levels and provide targeted teaching strategies. Comprehensive evaluation of students' overall quality will play a crucial role in the prediction model. Factors such as educational resources, teaching methods, student learning habits, and learning atmosphere in each college will affect the accuracy of the prediction. The full-coverage classification prediction of student performance using modern scientific and technological strategies holds great research significance for private universities' educational tasks and targeted teaching plans. [ABSTRACT FROM AUTHOR]

[Display omitted] Meeting climate targets requires robust carbon reduction strategies, particularly in the context of road transportation. This study presents a predictive model that integrates CO 2 emissions and operational costs for heavy-duty truck loads using Artificial Neural Networks (ANN) and Genetic Algorithms (GA) optimization. The model identifies the optimal vehicle driving profile by balancing environmental sustainability and economic efficiency. A strong correlation between vehicle weight and speed and CO 2 emissions was found, with the optimal weight and speed parameters being 49.67 tons and 31.00–36.61 km/h, respectively. The proposed model was tested across five scenarios, with the total cost per kilometer and emissions scenario yielding the best performance. The results demonstrate significant cost reductions, ranging from 31.4 % to 40.5 %, which not only reflect operational but also environmental cost savings. By optimizing driving parameters, fleet managers and decision makers can implement strategies to reduce operational and environmental costs, promoting sustainable transportation practices. [ABSTRACT FROM AUTHOR]

The driving characteristics and engine operating characteristics on vehicle carbon dioxide (CO 2) emissions of different types of vehicles are explored in this study. For Internal Combustion Engine Vehicles (ICEVs), vehicle specific power (VSP) is the parameter with the highest correlation coefficient with CO 2 emission rate, while for Hybrid Electric Vehicles (HEVs), it becomes engine speed. Due to the compound drive of fossil-fueled internal combustion engines and electric motors, the CO 2 emission rates of HEVs is no longer positive correlated with velocity-related vehicle dynamics presented by traditional VSP binning method. Therefore, a novel binary VSP binning model coupled with engine speed maps (VSP + M) is proposed to link the tailpipe emissions to vehicle activities and engine operating parameters. After well-designed configurations on the number of map divisions m and the number of elements into a tile z , the VSP + M model is able to achieve higher prediction accuracy along with better data usage. For HEVs, the prediction accuracy represented by R2 is observed over three-fold increase beyond 0.9, which embodies great value of binary model integrated with both externally observable variable (EOV) and internally observable variable (IOV) parameters in essence of the actual road traffic scenarios undergoing large-scale electrification. • The driving and engine operating characteristics on vehicle emissions are explored through RDE tests of HEVs and ICEVs. • According to the correlation analysis, the IVOs and EVOs that have strong correlations with CO 2 emissions are extracted. • A novel binary VSP model (VSP + M) is proposed to link the emissions to vehicle activities and engine operating parameters. • The VSP + M model is optimized by the values of m and z to improve the prediction accuracy of vehicle carbon emissions. [ABSTRACT FROM AUTHOR]

Action recognition, whose goal is identifying and extracting spatio-temporal features from video content, is a foundation of work in video understanding. However, current methods for sampling these features are computationally intensive and necessitate complex model architectures. In this paper, we improve 2D CNNs for action recognition tasks, keeping the model streamlined while making effective and complementary use of spatio-temporal features of videos. We propose a model, the Spatio-Temporal Analysis Network (STAN), which strikes a balance between model complexity and recognition accuracy. It contains two key components that affect spatio-temporal features: Temporal Embedding Head (TEH) and Spatio-Temporal Attention (STA). TEH introduces a differential analysis of actions, allowing the model to capture subtle temporal changes and enhance its representational capabilities. STA offers a novel perspective on video streams, improving the spatio-temporal representation without significantly increasing computational demands. It achieves this through a stylized spatial analysis of the features that differ from the conventional optical flow and depth map methods. The results in four datasets demonstrate our methodology's remarkable efficiency and accuracy. Compared to 3D CNNs, our method improves action recognition accuracy by 1.2% and reduces computational costs by 30%. With a dataset utilization rate of only 20% from UCF101, our model achieves an accuracy of 91.68%. • A single 2D CNN is implemented to achieve accurate action recognition and keep the model computational and parametric counts lean. • The combination of difference computation and residual connection refines fine-grained sampling of spatio-temporal features. • A new perspective on video stylized information to optimize the distribution of spatio-temporal features. • Effective differentiation of video action interval saliency and accurate identification driven by a small amount of data. [ABSTRACT FROM AUTHOR]

• Highly accurate and compact deep network for human stress recognition. • Integration of multiple design optimizations for edge device deployment. • Post-training quantization evaluated on Raspberry Pi Zero and Pi 4 devices. • Over 97 % accuracy achieved with a 22 KB model size across two datasets. • Foundational model applicable to diverse biomedical signal types. The effectiveness of deep learning algorithms in discerning stress levels from biomedical signals has attracted considerable attention. However, the computational demands of these algorithms often make them unsuitable for deployment on resource-limited devices. This paper introduces a highly accurate and compact deep residual network specifically designed for human stress recognition, optimized to operate with significantly reduced computational power, thereby facilitating deployment on edge devices. Its general-purpose architecture is intended to serve as a robust baseline model adaptable to various biomedical signal types, including photoplethysmography and accelerometer data. The network design integrates several optimization techniques, such as reduced kernel sizes, the substitution of pooling layers with strided convolutions, and the incorporation of inverted residual bottlenecks. Post-training quantization further enhances the model's efficiency, as validated on Raspberry Pi Zero and Pi 4 devices. Performance evaluation with two publicly available datasets demonstrated over 97 % accuracy while maintaining a compact size of 22 KB, effectively balancing accuracy, the number of parameters, and inference time. These results surpass many existing methods in accuracy while requiring substantially fewer computational resources, highlighting potential for integration into edge devices. [ABSTRACT FROM AUTHOR]

• Five models for predicting peak strength of sandstone are proposed. • The ABC algorithm can assist the hyper-parameters tuning of SVM effectively. • ABC-SVM has the best prediction performance by comparison. • The cosine amplitude method is used to analyze the correlation. • It provides new ideas for such prediction problems in rock engineering. The peak strength is a significant parameter in rock engineering, the traditional empirical strength criteria for rocks show good agreement with test results under specific conditions. However, it is not completely accurate for a wide range of loading stress domains and uncorrelated rock types. In this research, porosity, uniaxial compressive strength (UCS) and confining pressure are selected as input variables, and the artificial bee colony (ABC) algorithm is used to optimize the support vector machine (SVM) model. Finally, we validate and comparatively analyze the applicability of the models based on the testing set and the comprehensive evaluation indexes (namely correlation coefficient (R2), root mean square error (RMSE) and mean absolute percentage error (MAPE)). Meanwhile, the cosine amplitude method is applied to analyze the correlation between the peak strength and the input variables. The results indicate that both SVM model and ABC-SVM model are suitable for the prediction of peak strength under triaxial compression. Additionally, the ABC-SVM model obviously has better prediction performance by comparison. [ABSTRACT FROM AUTHOR]

Personalized federated learning (PFL) addresses the significant challenge of non-independent and identically distributed (non-IID) data across clients in federated learning (FL). Our proposed framework, "FairDPFL-SCS: Fair Dynamic Personalized Federated Learning with Strategic Client Selection", marks a notable advancement in this domain. By integrating dynamic learning rate adjustments and a strategic client selection mechanism, our approach effectively mitigates the challenges posed by non-IID data while enhancing model personalization, fairness, and efficiency. We evaluated FairDPFL-SCS using standard datasets, including MNIST, FashionMNIST, and SVHN, employing architectures like VGG and CNN. Our model achieved impressive results, attaining 99.04% accuracy on MNIST, 89.19% on FashionMNIST, and 90.9% on SVHN. These results represent a substantial improvement over existing methods, including a highest increase of 16.74% in accuracy on SVHN when compared to the best-performing benchmark methods. In particular, our method also demonstrated lower fairness variance, presenting the importance of fairness in model personalization, a frequently overlooked aspect in FL research. Through extensive experiments, we validate the superior performance of FairDPFL-SCS compared to benchmark PFL approaches, highlighting significant improvements over state-of-the-art methods. This work represents a promising step forward in the field of federated learning, offering a comprehensive solution to the challenges presented by non-IID data while prioritizing fairness and efficiency in model personalization. • A Fair Dynamic Personalized Federated Learning with Strategic Client Selection (FairDPFL-SCS) is proposed. • This research work provides a comprehensive analysis and categorization of related works with their pros and cons. • The FairDPFL-SCS model is a lightweight dynamic model which can be used for greater practicability. • This research work addresses non-IIDness issue in PFL with improved fairness and accuracy. [ABSTRACT FROM AUTHOR]

Process uncertainty has a significant impact on industrial image processing. Existing deep learning methods were established on high-quality datasets without considering the uncertainty. This paper proposes lightweight spatial-channel feature disentanglement modeling with confidence evaluation for uncertain industrial images. First, spatial-channel feature disentanglement modeling inspired by tensor decomposition aims to balance the computational efficiency and feature expression capabilities. Second, collaborative learning with confidence evaluation is designed to cope with uncertain samples. Then, representative features are fine-tuned on high-confidence datasets for optimal performance. Complexity analysis and experiments verified the effectiveness, computational efficiency, and robustness of the proposed model. • Lightweight spatial-channel modeling is proposed for the uncertain image. • Tensor decomposition balances model complexity and accuracy. • Collaborative learning with confidence evaluation improves the robustness. [ABSTRACT FROM AUTHOR]

The effective detection of decayed citrus fruit is challenging because infected fruit show few visual symptoms. A new method was proposed for early decayed orange detection using spectral visualization, data augmentation, and deep learning. First, the Borderline-Synthetic minority oversampling technique (SMOTE) algorithm was used to expand the decayed sample data, which solves the problem of a large gap in the number of samples from different categories (sound and decayed). Next, the spectral data was encoded into Gramian angle summation field (GASF) images, and the GASF image features were extracted using AlexNet-SVM for classification when the spectra of different samples were highly similar. Experimental results showed a classification accuracy greater than 95 %, full-wavelength features optimized, and the time for generating GASF images reduced by 59.43 % compared to the pre-optimization time. Thus, the proposed methodology provided a novel solution for defect detection in fruit and vegetables based on spectral analysis and indicated that combining one-dimensional spectral data with image processing using visual coding techniques is feasible. • Hyperspectral transmittance imaging was proposed for detection of decayed oranges. • Encoded one-dimensional transmittance spectra into visualized GASF images. • Borderline-SMOTE algorithm expanded small sample data. • An improved deep learning model based on spectral visual coding was constructed. • GASF-AlexNet-SVM is effective to classify decayed oranges with an accuracy of 95 %. [ABSTRACT FROM AUTHOR]

Energy loss due to non-uniform distribution of energy flux density on the standard compound parabolic concentrator (S-CPC) absorber, and the high processing cost of S-CPC curved reflector make it difficult to popularize. Based on replacing the curved reflector of the S-CPC with plane reflector, a quick optical simulation optimization method using particle swarm optimization (PSO) in full incident angles was proposed to construct multi-sectioned compound parabolic concentrator (M-CPC) to alleviate these problems. M-CPC models with different number plane reflectors that were designed by the PSO program, and its reliability was verified in a flexible laser experiment. The conducted analysis focuses on a detailed investigation of optical properties in optimized models. Among them, the average day beam radiation collected by M-CPC3 in Kunming equinox month was 3.1681 MJ exceeding that of S-CPC at 3.1631 MJ. In particular, the uniformity U of the energy flux density distribution on the absorber evaluated by combination with the numerical and spatial location distribution indicated that M-CPCs are more uniform than S-CPC. Moreover, it was analyzed that M-CPC consisting of multi-sectioned plane glass reflector may be more economical than S-CPC regarding cost consumption. [ABSTRACT FROM AUTHOR]

As a green, low-carbon and renewable energy, the geothermal energy has attracted great attentions. Abundant geothermal resources are hosted in several depleted oilfields. Geothermal development in abandoned oilfields has its advantages that operators are familiar with the geographic environment, geological conditions, and drilling and development techniques in the oilfield. In this study, a thermo-hydro- mechanical multi-field coupling mathematical model was established with data from Liubei Buried Hill reservoir in North China Oilfield. Then, the geothermal development in abandoned high-temperature oilfields was simulated, and the sensitivity of the geothermal development effect to well patterns, fracturing conditions, and working media was analyzed. The results show that the well pattern of one-injector and four-producer leads to the largest heat-exchange area. The hydraulic fracture model causes earlier thermal breakthrough but produces more heat than the non-fractured model. CO 2 has the better working fluid performance and leads to the better geothermal development effect than water. This study provides a theoretical basis for the high-efficiency development of geothermal energy in abandoned oilfields, and a guidance for research on geothermal development model and fracturing design. [Display omitted] [ABSTRACT FROM AUTHOR]

Vision transformers have achieved remarkable success in computer vision tasks by using multi-head self-attention modules to capture long-range dependencies within images. However, the high inference computation cost poses a new challenge. Several methods have been proposed to address this problem, mainly by slimming patches. In the inference stage, these methods classify patches into two classes, one to keep and the other to discard in multiple layers. This approach results in additional computation at every layer where patches are discarded, which hinders inference acceleration. In this study, we tackle the patch slimming problem from a different perspective by proposing a life regression module that determines the lifespan of each image patch in one go. During inference, the patch is discarded once the current layer index exceeds its life. Our proposed method avoids additional computation and parameters in multiple layers to enhance inference speed while maintaining competitive performance. Additionally, our approach 1 1 https://github.com/cjwcommuny/life-regression. requires fewer training epochs than other patch slimming methods. [ABSTRACT FROM AUTHOR]

Optimization research often confronts the challenge of developing time consuming processes. This article introduces an innovative approach that leverages the computational power of Graphics Processing Units (GPUs) to speed up that optimization process. We present an innovative adaptation of Particle Swarm Optimisation (PSO) to meet the requirements of multiobjective optimization problems. This approach aims to leverage the strengths of a multi-objective approach to perform energy consumption prediction using neural networks. By employing GPU parallel techniques, our method not only speeds up the optimization process but also enhances the efficiency of neural network training execution. The main advantage of our approach lies in its dual ability to simultaneously optimizing neural network architectures by determining the minimum number of hidden neurons and fitting the weights of the networks in order to achieve the lowest error. Preliminary results suggest a notable enhancement in prediction accuracy of forecasting electric energy consumption, as a result of optimizing the architecture and parameters of the neural network using the proposed method. This PSO adaptation stands out for its ability to address complex problems, increase efficiency and produce accurate predictions, making it a novel solution in Machine Learning heuristic methods for application in the solution of advanced prediction problems with time constraints from time series. • An adaptation of Particle Swarm Optimisation (PSO) is introduced for multiobjective optimization. • The GPU/CUDA parallel techniques accelerates the process of neural network execution. • The results show a significant improvement in prediction accuracy for energy consumption. • The adapted PSO method is highlighted for its ability generate accurate predictions. [ABSTRACT FROM AUTHOR]

Ensemble models in classification problems often encounter learning collision during joint training, where multiple base learners learn similar data representations concurrently. This phenomenon can diminish diversity and confidence in classification, especially in the context of medical image analysis, potentially leading to biased class predictions and classification errors. In this study, we tackle this issue by proposing ensemble models that combine both joint and independent training methodologies on the same medical image dataset. Our key contribution lies in explicitly controlling diversity through the design of the loss function. We fine-tuned the ResNet50V2 base learner, resulting in a significant 3% increase in training accuracy (86.00% compared to the previous 83.00%). Losses decreased from 0. 42 ± 0. 44 to 0. 40 ± 0. 42. For ResNet101V2, we observed a 2.27 percentage point increase in training accuracy (84.57% compared to the previous 82.30%) and reduced loss values to 0. 357 ± 0. 367 from the previous 0. 428 ± 0. 448. Furthermore, we conducted a comparative analysis of our optimized ensemble models, both with and without pruning, to assess their impact on model performance and to better understand their efficacy compared to earlier research work. The results underscore the effectiveness of our approach in mitigating learning collision and enhancing classification accuracy, particularly in the domain of medical image classification. Overall, our approach effectively reduces learning collision and improves classification accuracy as well as test accuracy on unseen medical images, addressing a significant gap in COVID-19 identification. This novel approach holds promise for ensemble models in medical image classification, particularly for lung-related diseases. • Collisions in learning affect predictions, especially in image analysis, leading to biases and errors. • Proposed controlling diversity through loss function for joint and independent training. • ResNet50V2 & ResNet101V2 loss decreased from 0. 42 ± 0. 44 to 0. 40 ± 0. 42 and from 0. 428 ± 0. 448 to 0. 357 ± 0. 367 , with 3% & 2.227% increase in accuracy respectively. • Evaluated optimized model against state-of-the-art, with equal class numbers in Section (4). • In summary, our method mitigates learning collisions in ensemble, yielding promising classification outcomes. [ABSTRACT FROM AUTHOR]

Personalized federated learning (PFL) is an exciting approach that allows machine learning (ML) models to be trained on diverse and decentralized sources of data, while maintaining client privacy and autonomy. However, PFL faces several challenges that can deteriorate the performance and effectiveness of the learning process. These challenges include data heterogeneity, communication overhead, model privacy, model drift, client heterogeneity, label noise and imbalance, federated optimization challenges, and client participation and engagement. To address these challenges, researchers are exploring innovative techniques and algorithms that can enable efficient and effective PFL. These techniques include several optimization algorithms. This research survey provides an overview of the challenges and motivations related to the model optimization strategies for PFL, as well as the state-of-the-art (SOTA) methods and algorithms which seek to provide solutions of these challenges. Overall, this survey can be a valuable resource for researchers who are interested in the emerging field of PFL as well as its potential for personalized machine learning in a federated environment. • A Comprehensive Review of Contemporary Approaches in Personalized Federated Learning (PFL). • Categorization and classification of various methods and solutions to provide comprehensive taxonomies. • A diverse array of literature that addresses emerging challenges and cutting-edge solutions in the field. • This paper outlines and proposes new research directions that can be pursued by the interested community. [ABSTRACT FROM AUTHOR]

With the development of industrialization and rapid urbanization, more and more rural resources are being directed into urban spaces, leading to rural poverty. The gaps between urban and rural have gradually increased alongside serious "rural diseases". Thus, rural revitalization is an essential and important strategy in the new era for realizing better urban-rural integration. This paper proposes a rural system evaluation model to divide rural development spatial-multivariate zones as evidence for exploring and optimizing rural sustainable development models. The result shows that: in 2015, Miyun District was divided into six zones, including a city zone, a town development zone, an industrial zone, an agricultural zone, a leisure zone and an ecological zone. The agricultural zone, leisure zone and ecological zone showed evidence of village hollowing and waste, a weakening of agriculture and poorer infrastructure, leading to a lower-level economy when compared to other zones. Thus, this study explores a revitalization path for Miyun District through four methods: mechanism, skilled workers, industry and technology, and proposes three optimizing models to solve rural problems. Building a new town development zone, developing multi-talent education and integrating first-second-third industries would reduce the gap between the urban and rural and would realize urban-rural integration more rapidly. • It built a rural system evaluation model with many factors weighting. • It divided the spatial-multivariate by the value of rural development. • It proposed rural revitalization ways and optimizing models to solve rural problems. [ABSTRACT FROM AUTHOR]

This research looks into children's drawing strategies that focus on sequencing and order of strokes for children to produce a seriation object. The drawing strategies were examined according to 6 sets of logical structures that are; (1) embedding; (2) accretion stacking; (3) anticipated embedding; (4) anticipated stacking; (5) partial framing; and (6) full framing. Past work studied these logical structures for drawings on paper and used the traditional method of observation for evaluation. This traditional method is an exhaustive approach and leads to in-accuracies due to human error as a result of ambigous data. To solve this, we extend the work for drawings on touch screen where children's drawing data were quantified using a novel deep learning hybrid model (Fuzzy string matching optimized with Levenshtein Distance in LTSM - FLSTM) to classify the drawn strategies. We developed a touch drawing application with 8 seriation objects as the drawing task. 32 children of age between 5 and 12 years old took part in this study with a total of 420 drawings collected. Comparative model performance was done between the proposed novel model with existing models such as Long Short-term Memory model (LSTM), Convolution Neural Network model (CNN) and Fuzzy-CNN model for comparison in drawing classification accuracy. The results showed that the proposed novel deep learning hybrid model outperformed other models with a precision score of 89.1%, recall of 88.6% and F1 score of 88.6%. With assistance of the proposed deep learning model, we were able to explore and understand more about human psychological behaviour through the developed children drawing system. [ABSTRACT FROM AUTHOR]

Low computational cost artificial intelligence (AI) models are vital in promoting the accessibility of real-time medical services in underdeveloped areas. The recent Once-For-All (OFA) network (without retraining) can directly produce a set of sub-network designs with Progressive Shrinking (PS) algorithm; however, the training resource and time inefficiency downfalls are apparent in this method. In this paper, we propose a new OFA training algorithm, namely the Progressive Expansion (ProX) to train the medical image analysis model. It is a reversed paradigm to PS, where technically we train the OFA network from the minimum configuration and gradually expand the training to support larger configurations. Empirical results showed that the proposed paradigm could reduce training time up to 68%; while still being able to produce sub-networks that have either similar or better accuracy compared to those trained with OFA-PS on ROCT (classification), BRATS and Hippocampus (3D-segmentation) public medical datasets. The code implementation for this paper is accessible at: https://github.com/shin-wl/ProX-OFA. [ABSTRACT FROM AUTHOR]

Lycopene content is one of the most important indicators for tomato quality evaluation. Traditional detection methods are destructive and time-consuming. This study firstly used the multi-point full transmission visible and near-infrared spectroscopy for online detection of lycopene in two cultivars of tomatoes ('Provence' and 'Jingcai No.8′). The weighted average was applied to process multi-point spectral data. Two orientations (O1 and O2) and three preprocessing methods were considered and least angle regression (LARS) with L1 and L2 norms was used for wavelength selection. The independent partial least squares regression (PLSR) model was established. The PLSR approach combined with LARS-L1 and O2 yielded the best lycopene prediction for 'Provence' tomato (Rp = 0.96, RMSEP = 13.44 mg kg−1) and 'Jingcai No.8′ tomato (Rp = 0.95, RMSEP = 7.43 mg kg−1). As an extension, a general model was also established and proved its feasibility. This study provides a novel methodology for accurate and rapid detection of lycopene in tomatoes. • Online detection of lycopene in the two varieties of tomatoes was studied. • Multi-point full transmission Vis-NIR spectroscopy was firstly proposed. • The weighted average method improves the quality of multi-point spectra data. • The optimal detection orientation and characteristic wavelengths were determined. • The developed independent/general models were effective for predicting lycopene. [ABSTRACT FROM AUTHOR]

• Open-source dataset: Five categories of dataset exported using multiple augmentation techniques. • Single-stage model that achieves weed detection and segmentation. • Integrating C3x module within the backbone for detailed feature extraction. • Model trained on C4 category achieved the best detection and segmentation scores of 85.4 % and 82.1 %, respectively. • Model with ONNX format gained 1.25x inference speed on an edge device. Deep learning (DL) inspired models have achieved tremendous success in locating target weed species through bounding-box approach (single-stage models) or pixel-wise semantic segmentation (two-stage models), but not both. Therefore, the goal of this research study was to develop a single-stage DL architecture that not only locate weed presence through bounding-boxes but also achieves pixel-wise instance segmentation on unmanned aerial system (UAS) acquired remote sensing images. Moreover, the developed architecture experiments on integrating a novel C3 and C3x module within its backbone for dense feature extraction, as well as ProtoNet (Prototypical network) in its head component for weed masking. Furthermore, the proposed architecture has been trained on five categories of dataset exported using multiple combinations of various dataset augmentation techniques, namely, C1, C2, C3, C4, and C5, for which multiple metrics were assessed on desktop graphical processing unit (GPU) and a palm-sized edge device (AGX Xavier). Results suggest that category C4, a combination of six data augmentation techniques, outperformed the remaining categories by achieving precision scores of 85.4 % (bounding-boxes) and 82.8 % (masking) on a GPU. Whereas, the same model converted to TorchScript was able to achieve 79.1 % and 77 % bounding-box and masking accuracy on an edge device, respectively. The model developed in this research has two potential applications when integrated with site-specific weed management technologies. First, it enables real-time weed detection, allowing for the immediate identification of weeds for spot-spraying applications. Second, it facilitates instance weed masking, aiding in the estimation of weed growth extent in actual field conditions. Moreover, the developed architecture combines both computer vision applications - detection and instance segmentation – to provide comprehensive information about weed growth, eliminating the need for multiple algorithm. [ABSTRACT FROM AUTHOR]

N 2 injection to enhance coal seam gas drainage is an effective technology for coalbed methane recovery. Nevertheless, there is a limited body of research addressing the application of ECBM technology for gas management in low permeability coal seams within underground coal mines, and the effect is uncertain. Additionally, numerical simulators have been challenged to accurately predict methane emissions during gas injection and replacement. In this study, firstly, an underground field trials of N 2 injection replacement in a coal mine with low-permeability coal seams in China were conducted, and a comparative analysis of "N 2 injection to enhance gas drainage" and "conventional drainage" technique was performed. The result show that injecting pressurized N 2 into a low-permeability coal seam could increase the pressure gradient to promote the directional flow of free gas from the coal seam cracks to the gas-production boreholes. After N 2 injection, the mixed flow rate and CH4 flow rate significantly increased. Additionally, the gas content in the coal seam in the experimental area significantly decreased. Secondly, the lag-effect was discussed. Upon discontinuation of N 2 injection, the gas-producing mixed flow initially increased slightly, then gradually decreased, and eventually settled at a high level. Despite the attenuation of N 2 injection volume, the CH 4 flow rate continued to increase during intermittent injection and remained high after gas injection discontinuation. Finally, three deep learning algorithms (BP, LSTM, and TCN) were employed to predict the dynamic change of gas drainage in coal seams displaced by N 2 injection, based on the results of the field test. The results indicated that deep learning algorithms exhibited superior prediction performance. The TCN model demonstrated the lowest approximation error rate, displaying optimal performance in predicting the CH 4 flow rate during nitrogen injection for enhanced gas drainage. Additionally, it exhibited good applicability to other similar field test results. [Display omitted] • A field test of N 2 -ECBM in low-permeability coal seams was carried. • The gas production laws of "N 2 injection to enhance drainage" were analyzed. • The lag-effect during intermittent gas injection was analyzed. • 4.The deep learning algorithms was applied to "N 2 injection to enhance drainage". [ABSTRACT FROM AUTHOR]

Established an effective relationship between catalyst characteristics and BTEXN. [Display omitted] • Synergistic catalysis of 1-naphthol by coupling macroporous γ-Al 2 O 3 and HZSM-5. • Machine learning modeling is used to optimize the loading type and mass ratio. • 1-naphthol is mainly converted to naphthalene on γ-Al 2 O 3 , and then to BTEX on HZSM-5. High energy consumption, environmental pollution, and poor conversion are major challenges to the efficient utilization of coal tar. Tandem catalysis is considered one of the most promising technologies for the high-value utilization of coal tar. The main challenge for the efficient utilization of coal tar is the low added value and complexity of the products, while tandem catalytic technology is considered as one of the most promising technologies for the high-value utilization of coal tar. It is report here a strategy for the highly selective conversion of 1-naphthol to light aromatics catalyzed by a homemade tandem of macroporous γ-Al 2 O 3 and HZSM-5 catalysts with large specific surface area. We demonstrate that 1-naphthol first undergoes dehydroxylation on γ-Al 2 O 3 , and then the naphthalene-dominated intermediates undergo further cracking-selective reactions on the adjacent HZSM-5 bed to produce BTEX. This work was carried out under the loading conditions optimized by machine learning modeling. By adjusting the loading mode, adjusting the mass ratio (bed thickness), optimizing the reaction temperature, enhancing the proximity effect, adjusting the pore structure of HZSM-5, and modifying HZSM-5 with three metals 1Ni-5W-0.2La, the maximum conversion rate of 1-naphthol (R 1-naphthol) and the selectivity of light aromatics BTEXN (S BTEXN) reached 67.31 % and 47.43 %, respectively. Additionally, the structure–activity relationship between the physicochemical properties of the catalyst and the S BTEXN , R 1-naphthol was also explored. The results of this study can provide a unique reference perspective for the efficient preparation of light aromatic hydrocarbons from phenol-rich coal tar. [ABSTRACT FROM AUTHOR]

Enabling deep neural networks for tight resource constraint environments like mobile phones and cameras is the current need. The existing availability in the form of optimized architectures like Squeeze Net, MobileNet etc., are devised to serve the purpose by utilizing the parameter friendly operations and architectures, such as point-wise convolution, bottleneck layer etc. This work focuses on optimizing the number of floating point operations involved in inference through an already compressed deep learning architecture. The optimization is performed by utilizing the advantage of residual connections in a macroscopic way. This paper proposes novel connection on top of the deep learning architecture whose idea is to locate the blocks of a pretrained network which have relatively lesser knowledge quotient and then bypassing those blocks by an intelligent skip connection, named here as Shunt connection. The proposed method helps in replacing the high computational blocks by computation friendly shunt connection. In a given architecture, up to two vulnerable locations are selected where 6 contiguous blocks are selected and skipped at the first location and 2 contiguous blocks are selected and skipped at the second location, leveraging 2 shunt connections. The proposed connection is used over state-of-the-art MobileNet-V2 architecture and manifests two cases, which lead from 33.5% reduction in flops (one connection) up to 43.6% reduction in flops (two connections) with minimal impact on accuracy. [ABSTRACT FROM AUTHOR]

Photoperiod and light intensity have been shown to play important roles in the growth of larval fish. In terms of the effect of light conditions on the larval growth of Nile tilapia, previous studies are confined to examining the effect of only photoperiod rather than of their combined effects. The face-centered composite experimental design and response surface methodology were utilized to investigate the combined influences of photoperiod and light intensity on the feed conversion and larval growth of Nile tilapia in this study. Based on the results of pilot experiments, photoperiod ranged from 8 h to 24 h, light intensity from 100 lx to 2000 lx (1.26–25.20 μmol m−2 s−1). Results showed that under indoor recirculating rearing system varying growth rhythms occurred with different light conditions; interaction between the two factors was detected, and the light intensity was more important in impacting on growth than photoperiod in the larval stage, implying that light intensity and photoperiod ought to be examined in concert rather than separately. If light intensities are held invariable, the photoperiod studies would lead to the choice of >18 h photoperiod for the best larval growth. Response surface of the larval growth was roughly consistent with that of feed conversion, indicating that the larval growth was realized through feed conversion. Relationship of the larval growth and feed conversion to artificial light conditions could be reliably quantified in a second-order form, through optimization of which optimal two-factor combination, 14–16 h/1650–1900 lx (20.79–23.94 μmol m−2 s−1), was derived with the reliability as high as ca. 98%. There is therefore the potential for maximizing larval growth in hatcheries via manipulation of artificial light regimes. According to our findings, continuous lighting should not be recommended for optimal growth in larviculture. These findings underscore the important role of light conditions during the larval stage of Nile tilapia and should be taken into consideration for the optimization of rearing protocols in Nile tilapia hatcheries. • Larval growth of Nile tilapia can be manipulated by artificial light regimes. • Larval growth is realized through increased feed conversion efficiency. • Light intensity is more important than photoperiod in affecting larval growth, their impacts should be investigated together. • Long photoperiod (including continuous lighting) is not recommended for the larval growth. • Optimal combination of light regimes is determined via simultaneous optimization of reliable models of two responses. [ABSTRACT FROM AUTHOR]

Abstract Pinctada martensii is one of several important pearl oysters. Ammonia is typically added to the seawater and used during artificial reproduction to improve the fertilization rate. However, the optimal ammonia concentration has not been determined. To determine the optimal ammonia concentration for fertilization success, the effect of the ammonia concentration on the fertilization rate in P. martensii was investigated under the simultaneous influence of temperature and salinity. A Box-Behnken design and response surface method were used in this study. The results showed that the ammonia concentration had a significant effect, which varied quadratically with the ammonia concentration in the ranges of the factor regimes (P <.01). The interactions between the ammonia concentration and the temperature/salinity had a significant effect (P <.05), indicating that the effect of the ammonia concentration depended on the combination of temperature and salinity; therefore, their combined effects should be taken into account when determining the optimal ammonia concentration. The effect of the ammonia concentration was less than that of the temperature but more than that of the salinity. A reliable model of the fertilization rate using the three factors was constructed. The results of the model optimization demonstrated that a fertilization rate of 96.16% was achieved at an ammonia concentration of 1.85 mmol L−1 and a temperature-salinity combination of 27 °C/32 practical salinity unit (psu). The ideal fertilization rate was verified experimentally using the optimal ammonia concentration while maintaining the temperature and salinity at natural levels during the spawning season. The results of this study can be used to refine the seed production efficiency of the pearl oyster. Highlights • Effect of ammonia on fertilization in Pinctada martensii under concurrent influence of temperature and salinity investigated. • Modifying effects of temperature and salinity should be considered upon determining optimal ammonia concentration. • Optimal ammonia concentration determined via optimization of reliable fertilization model. • Optimization results verified. [ABSTRACT FROM AUTHOR]

• A light deep framework is constructed for fault diagnosis of a piston pump. • Multi-sensor and multiple channel signals are monitored for information mining. • Adaptive learning of model hyperparameters is achieved by Bayesian algorithm. • The design of batch normalization can improve the stability of the model. • The proposed diagnosis method can accurately achieve the fault identification. The health condition of hydraulic axial piston pumps is crucial for the safety and reliability of hydraulic transmission systems. Diagnosis results of traditional methods indicate the high reliance on the experience, and current deep model-based methods are confronted with the difficulty of parameter tuning. A light adaptive deep framework is therefore constructed to reduce reliance of diagnosis on expert experience and still achieve the automatic screening of model hyperparameters. First, multi-sensor and multiple-channel signals of a piston pump are acquired for comprehensive raw data input. The raw one-dimensional signals are then converted into two-dimensional images using continuous wavelet transform. Then, a light deep model is built based on the convolutional operation and batch normalization techniques. The final model is obtained via global optimization of Bayesian algorithm. Next, the improved deep model is adopted for failure recognition of the essential components in the piston pump based on the transformed images. The average diagnosis accuracy achieves 97.46%, 98.71%, and 99.94% based on vibration signal, acoustic signal, and pressure signal respectively. The results reveal that the typical fault of the piston pump can be recognized intelligently and accurately with the proposed diagnosis method. [ABSTRACT FROM AUTHOR]

• Open-source dataset: 3929 images and 12 k bounding-box annotations used in this study. • Deep learning model optimization using pruning and integrating re-parameterization module. • Assessing the effects of training multiple image resolution on the optimized YOLO-Spot model. • YOLO-Spot_M model achieves accuracy (+1.3 %) and mAP (+2.7 %) compared to YOLO-base model. • YOLO-Spot_M with half-precision gains 5X times inferencing speed on an edge device. Recent advancements in deep learning (DL)-based model optimization techniques have resulted in better weed identification accuracy. However, optimizing these models to identify weeds in images captured using small unmanned aerial system (UAS) has not been much explored. Moreover, leveraging the optimized model on resource constrained edge platform that could be easily integrated with UAS for real-time weed identification could be of significant advantage in developing precision aerial spraying weed management technology. Therefore, this study proposes YOLO-Spot model that is based on YOLOv7-tiny architecture, has been optimized and reconstructed to identify weeds amongst crop plants in aerial images and videos. The optimized model tends to use a smaller number of trainable parameters and reduced feature map sizes for weed identification. Most of the redundant convolutional layers along with feature map sizes have been reduced coupled with an integration of a novel module re-parameterized convolutional layer (RCL) within the neck component of the network. Furthermore, YOLO-Spot model has been trained on the three image resolutions, 320 × 320, 640 × 640 and 1280 × 1280, and has been named as YOLO-Spot_S, YOLO-Spot_M and YOLO-Spot_L, respectively. Out of all the variants, YOLO-Spot_M model has achieved significant prediction accuracy as compared to other variants and a denser layered model YOLOv7-Base. YOLO-Spot_M model utilizes over 75 % less parameters and 86 % reduced GFLOPs compared to YOLOv7-Base. As per the results, YOLO-Spot_M has outperformed YOLOv7-Base by achieving +1.3 % and +2.7 % overall accuracy and mAP(@0.5), respectively. The optimized architecture utilizes 4X less power (in W) when trained on a normal graphical processing unit (GPU). Moreover, converting YOLO-Spot_M to half-precision (FP16) for resource constrained device deployment (AGX Xavier), led to a +0.6 % accuracy and 5X faster weed recognition accuracy in aerial images and videos during inferencing. Based on the metrics obtained, YOLO-Spot_M model is recommended model that could be integrated with remote sensing technologies for site-specific weed management. [ABSTRACT FROM AUTHOR]

The prediction accuracy of chemical kinetics models can be improved efficiently by using automatic model optimization techniques. In the optimization, an objective function, which quantifies the differences between model responses and experimental data for quantities of interest, is minimized by calibrating the reaction rate parameters of a model within their uncertainty limits. Consequently, the values of the model predictions become closer to those of the measurements. Typically, a point-wise objective function, which is based on function components separately for each measurement over the investigated domain, is used in the model optimization. Quantities of interest are often functions of various physical condition parameters, such as temperature, pressure, and equivalence ratio. However, the point-wise objective function does not consider the correlation between data and their corresponding physical conditions. Thus, in this work, a new objective function is proposed, which uses a surface-matching (SM) method. It evaluates the similarity between surface shapes of the predicted and measured values, which is quantified in form of two user-defined physical condition parameters. By minimizing this function, the joint dependence of model predictions on physical conditions is optimized in conjunction with the point-wise model prediction accuracy. A chemical mechanism of oxymethylene ethers is optimized in this work as an example. The model is calibrated with the point-wise, curve-matching (CM)-based, and SM-based objective functions. The optimized models are compared and the results are discussed. It is shown that the optimization with the SM-based objective function yields improved results for certain cases compared to using the point-wise objective function. This model also provides the best prediction accuracy in terms of joint physical condition dependence. In addition, a better overall performance is achieved by adjusting the ratios between the component functions in the objective function, which demonstrates that the definition of objective functions plays a crucial role for model optimization. [ABSTRACT FROM AUTHOR]

Accurate electricity price forecasting (EPF) is crucial to participants and decision-makers within the electricity market. This paper reviews 62 screened literature works on EPF during 2012–2022 in terms of model structure and determinants of electricity price and discusses the evaluation process, model type, research sample, and prediction horizon. From the above efforts, we find that (1) data preprocessing and model optimization are often used to improve forecasting model accuracy; while performance evaluation is essential, extensive performance evaluation benchmarking is still missing; (2) considering electricity price determinants can significantly improve forecasting model accuracy, but there is disagreement over how many and which determinants should be accounted for; (3) while most existing research focuses on point forecasting, interval and density forecasting are more responsive to the range and uncertainty of electricity price changes. [Display omitted] • State-of-the-art electricity price forecasting modeling techniques are reviewed. • Data preprocessing and model optimization are often used to improve accuracy. • Dual decomposition strategy develops the data preprocessing methods. • Considering electricity price determinants can improve forecasting performance. • Interval and density forecasts are more responsive to electricity price changes. [ABSTRACT FROM AUTHOR]

Stacking ensemble learning is one of the most effective integration technologies and is increasingly applied to near-infrared spectroscopy combined with chemometrics methods. The prediction accuracy of Stacking is primarily affected by the selection of different models. However, many current studies are mainly artificial selection models' combinations. It affects the model's prediction accuracy and increases the algorithm's difficulty. It is difficult to efficiently and accurately find the optimal configuration scheme. This study applies a genetic algorithm to find the optimal base and meta learner combinations in Stacking ensemble learning. This method uses the near-infrared spectral data set of corn seed germination rate. First, select the best pretreatment methods for seven models, including Gaussian process regression (GPR), SVR, PLS, etc. The above seven single learners after pretreatment are taken as the candidate base learner, and then random forest (RF), SVR, PLS, and GPR are taken as the potential meta learner; use a genetic algorithm to select the optimal model combination configuration and generate GA-Stacking algorithm. The model prediction results of the improved model GA-Stacking are compared with several single models and Stacking ensemble learning via the artificial selection model combinations. The results show that the prediction performance using the GA-Stacking ensemble learning model is optimal, R2 is 0.9022, and RMSE is 0.1100. The experiment shows that the model combination selected by the genetic algorithm has significantly improved the prediction performance of the Stacking ensemble learning model and reduced the risk of the model's overfitting. • Genetic algorithm is used to optimize model combinations for Stacking ensemble learning. • Overcome the disadvantage of artificial selection of stacking ensemble learning model combination optimization. • Verify the effectiveness of the Stacking ensemble learning optimization algorithm. [ABSTRACT FROM AUTHOR]

Abstract With the growing desire for "lucid waters and lush mountains", energy savings and emission reductions of automobiles have received unprecedented attention. As a result, how to achieve economical driving is also a focus of current research. The engine dynamic fuel consumption model is the basis for research on automobile fuel economy. To overcome the deficiency of the original transient fuel consumption models based on "steady-state prediction plus transient correction", a new model called BIT-TFCM-3 was developed. The new model was verified using measured fuel consumption data from Argonne National Laboratory. The results show that, compared with the original models, the new model not only has a higher computing speed but also a significantly improved prediction accuracy. Highlights • The deficiencies of the original fuel consumption models are analyzed. • Block interpolation can increase the interpolation speed. • A new model based on "steady-state prediction plus transient correction" is developed. • The measured fuel consumption data from Argonne National Laboratory is used to verify the model. • The model can be used as an economic evaluation tool for new vehicle technologies. [ABSTRACT FROM AUTHOR]

Highlights • The NARX model and Ali-Schaeffer model were compared at individual cow level. • Six input treatments including parity weight combinations were tested and compared. • The NARX Model was more accurate than the Ali and Schaeffer model. • The effectiveness of the parity weight treatment varied between cow groups. • Parity weight trends were a determining factor in the success of the treatments. Abstract The objectives of this study were to compare the prediction accuracy of two milk prediction models at the individual cow level and to develop, compare and evaluate six input data preprocessing treatments designed to factor parity information into the milk prediction model configuration process. The two models were a nonlinear auto-regressive model with exogenous input and a polynomial curve fitting model. These were tested using six different parity data input treatments. Different combinations of static parity weight, dynamic parity weight and removal of the first lactation data were selected as input treatments. Lactation data from 39 individual cows were extracted from a sample herd of pasture-based Holstein-Friesian cattle located in the south of Ireland and situated in close proximity. The models were trained using three years of historical milk production data and were employed for the prediction of the total daily milk yield of the fourth lactation for each individual cow using a 305-day forecast horizon. The nonlinear auto-regressive model with exogenous input was found to provide higher prediction accuracy than the polynomial curve fitting model for individual cows using each input treatment. An improvement in forecast accuracy was observed in 62% of test cows (24 of 39). However, on average across the entire population, only part of the treatments delivered an increase in accuracy and the success rate varied between test groups. Prediction performance was strongly influenced by the cows' historical milk production relative to parity and also the prediction year. These results highlighted the importance of examining the accuracy of milk prediction models and model training strategies across multiple time horizons. Removal of the first lactation and applying static parity weight were shown to be the two most successful input treatments. The results showed that historical parity weighting trends had a substantial effect on the success rate of the treatments for both milk production forecast models. [ABSTRACT FROM AUTHOR]

Abstract The joint effects of temperature (18–30 °C) and photoperiod (0–24 h) on yolk utilization in the yolk-sac larvae of Pelteobagrus ussuriensis were investigated under laboratory conditions using a central composite design (face-centered) and response surface methodology. The results indicated that a suitable environmental temperature and photoperiod significantly increased yolk utilization. In contrast, unsuitable environments (e.g., low temperatures and complete darkness; low temperatures and continuous lighting; high temperatures and complete darkness; and high temperatures and continuous lighting) significantly reduced yolk utilization, albeit to varying degrees. Temperature affected yolk utilization independently of photoperiod, and was more important than photoperiod. Yolk utilization was synchronous with larval growth, showing the conversion of vitelline nutrients into larval growth. Predicated upon the reliable second-order relationship of yolk utilization with temperature and photoperiod, the optimal temperature-photoperiod combination was derived via model optimization and verified. The application of our results may help to increase the efficiency of seed production in the larviculture of this species. Highlights • Optimal rearing environment of temperature and photoperiod was attained through optimizing reliable models and verified. • Temperature and photoperiod influenced vitelline utilization in a non-interactive fashion. • Temperature was more important than photoperiod in impacting on vitelline utilization. • Variations of yolk utilization with different environments were in synchrony with larval growth of Pelteobagrus ussuriensis. [ABSTRACT FROM AUTHOR]

Abstract Designing reasonable departure schedule is the key step to realize the urban transit priority. It can not only reduce the operating cost of bus company, but also guarantee convenience for passengers. This paper estimates the travel time between bus stations based on the historical trajectory data of the bus, and then combines the number of passengers get on and off at each station to optimize the departure timetable. In addition, several constraints including actual travel time, limited capacity and arrival time distribution type are considered in the optimization models to effectively and comprehensively estimate the passenger waiting time. Finally, a hybrid algorithm combining Genetic Algorithm (GA) and Simulated Annealing Algorithm (SAA) is proposed to search optimal solution in scheduling model. A case study is applied to testify the effectiveness of proposed models. In the experiments, we compare optimization results of proposed method to traditional genetic algorithms, and the results show the superiority and feasibility of the hybrid optimization approach. Highlights • We apply Poisson distribution to determine the passengers waiting time. • An optimization model considering actual travel time, limited capacity and arrival time is proposed. • We propose a hybrid algorithm to generate departure timetable. • The results prove the superiority and feasibility of the proposed optimization method. • We discuss the impact of vehicle resource constraints on the scheduling scheme. [ABSTRACT FROM AUTHOR]

Chemometric methods permit the construction of classifiers that effectively assist in monitoring safety, quality and authenticity of meat based on the near-infrared (NIR) spectral fingerprints. Discriminant techniques are often considered in multivariate quality control. However, when the authenticity of meat products is the primary concern, they often lead to an incorrect recognition of new samples. The performances of two class modeling techniques (CMT) in order to recognize meat sample species based on their NIR spectra was compared – a one-class classifier variant of the partial least squares method (OCPLS) and the soft independent modeling of class analogy (SIMCA). Based on obtained sensitivity and specificity values, OCPLS and SIMCA can be considered as an effective CMT for the classification of complex natural samples such as studied meat samples (with a relatively large variability). Moreover, particular attention was paid to the optimization and validation of a one-class classification model. [ABSTRACT FROM AUTHOR]

Evaporation technology has unique advantages in the enrichment of radioactive ions, which can deal with the high concentration of radioactive wastewater. A integrated mechanical vapor compression (MVC) enrichment system with high efficiency filter was proposed and studied experimentally for radioactive wastewater, whose decontamination factor was 37107.1 for Co2+, 2741.3 for Sr2+, and 52545.5 for Cs+, which saved energy by 76.7–84.3% and operation cost by 72.0–81.0% when compared to the conventional single effect (SE) evaporation system. Exergy destruction and efficiency of the single screw compressor were 7.26 kW and 34.0%, respectively, which should be prioritized for improvement. Mathematical models based on both the heat transfer coefficient (HTC) and coefficient of performance (COP) were established and optimized experimentally, the ideal running parameters were 90–110 kPa for evaporating pressure, 8.5–11.5 °C for heat transfer temperature difference, 1.50–1.60 for compression ratio and 2.0–4.0 kPa for filtration pressure difference, respectively. It will supply the theoretical and data basis for development and design of MVC enrichment system of radioactive wastewater. • A MVC enrichment system for radioactive wastewater is proposed and built. • MVC greatly saves energy compared to the conventional single-effect system. • Compressor should be prioritized for improvement according to exergy analysis. • Mathematical models based on the HTC and COP are established and validated. • Running parameters of MVC are optimized based on the developed models. [ABSTRACT FROM AUTHOR]

Ensuring effective drinking water disinfection, remaining a certain amount of residual chlorine, and controlling disinfection by-product formation were very important for guarantying water quality safety and protecting public health; thus, the chlorine decay model and corresponding disinfection by-product formation model were necessary. This paper proposed a mixed-order chlorine bulk decay model (two parameters) based on Taylor's formula and derived its analytical solution. The accuracy of the mixed-order model was evaluated by comparing it with the nth-order model. To optimize the model and reduce the number of parameters required to be calibrated, the relationship of parameters with temperature, initial chlorine concentration, TOC and inorganic substance (ammonia nitrogen and iodide ion) was explored. The result proved that one of the parameters could be regarded as temperature dependent only. Meanwhile, the temperature equation of the model parameters was established by the Arrhenius formula. Subsequently, this paper selected trihalomethane as the target and study the linear relationship between chlorine consumption and trihalomethane formation. The results indicated that the liner slope had little correlation with initial chlorine concentration and temperature. On this basis, the corresponding trihalomethane model was built and its performance was proven to be good. The modeling developed in this work could be applied to drinking water distribution systems for residual chlorine and trihalomethane prediction, and provided a reference for the decision involving water quality. [Display omitted] • 1. Propose a mixed-order model with an analytical solution and fewer parameters. • 2. Parameter K was temperature-dependent and little related to ICC, TOC, NH 4 +, I−. • Linear relationship between chlorine consumption and trihalomethane was studied. • Establish a corresponding rihalomethane generation model and verify its accuracy. [ABSTRACT FROM AUTHOR]

• A three-dimensional molecular model of middle-rank bituminous coal is carefully constructed. • The reliability of the model was verified by FTIR, NMR, density and MEP methods. • The energy level orbital and electrostatic potential of the model is investigated. A suitable model can reflect the properties of coal well. In this study, the properties of HongLiuLin (HLL) coal were studied through industrial analysis, XPS, 13C NMR and FTIR. The results show the coal has middle-rank bituminous coal characteristics, and the ratio of aromatic bridge carbon to aromatic peripheral carbon is 0.16. Benzene and naphthalene ring are the main components of the aromatic structure. N and S exist in pyrrole nitrogen and fatty sulfur, respectively. The hypothesis method was used to deduce the molecular formula C 130 H 73 O 21 NS of the coal model. The stability of the carbon skeleton was improved through the Van Der Waals (VDW) forces in non-bonding energy. Verification of NMR, FTIR and density was completed based on quantum chemistry and molecular mechanics. The coal model energy level orbitals analysis shows that the LUMO and HOMO orbitals are located in the aromatic functional groups. This work provides a reference for further research on the nature of coal in north Shaanxi and improving the utilization rate of coal. [ABSTRACT FROM AUTHOR]

Climate essentially controls the supply of ecosystems, species ranges, and process rates on Earth. Modeling hydrological processes for hilly catchments dominated by snow cover and glaciers is complex and relies on improved calibration and uncertainty analysis methods to standardize the watershed based ecosystem management practices. In this study, a modified curve number (CN) approach has been employed to simulate streamflow and water yield at sub-catchment scale over hundred years utilizing Soil and Water Assessment Tool (SWAT), which relies on the main physical factors of the ecosystem such as landuse/landcover and soil. The model calibration and validation strength was evaluated using coefficients of determination (R 2 ) and Nash-Sutcliffe Equation (NSE) objective functions. The uncertainties (percentage) occurred in the modeled streamflow were estimated using two optimization algorithms such as the Sequential Uncertainty Parameter Fitting Approach (SUFI2) and the Parametric Solution (ParaSol). A parameterization based sensitivity analysis was carried-out to recognize the most influencing model calibration parameters. Statistical downscaling of daily temperature and precipitation datasets was performed utilizing Coupled Model Intercomparison Phase Five (CMIP5) Global Circulation Models (GCMs) with their Representative Concentration Pathway (RCP) experiments. The downscaled temperature and precipitation were utilized to assess the climate change impact on streamflows at sub-catchment scale. The historical and projected scenarios of streamflow (at the outlet) and water yield (at sub-catchment scale) showed substantial variabilities in their amount in both temporal (1991–2100) and spatial scales (sub-catchment 1 to sub-catchment 7). The magnitude of change analysis confirmed a substantial increase in the water yield across all the sub-catchments over Himalaya. The percent of change analysis ensured that the magnitude of change of water yield is highly vulnerable in the Himalayan catchments. The GCMs based projected scenarios demonstrated a consistent increase in the streamflow at both outlets (e.g. Lachung and Chungthang). The overall results show a consistent increase in precipitation and water yield amount over the Himalayan catchments. The variable streamflow in terms of amount and intensity may disrupt the ecosystem services in the Himalayan catchments. [ABSTRACT FROM AUTHOR]

Building integrated photovoltaic thermoelectric (BIPVTE) wall system is highly energy efficient and self-adaptive to the environment. This sophisticated system is supported by the co-work of PV module for solar radiation transformation, air gap for thermal dissipation and thermoelectric radiant panel system (TERP) for active radiant cooling/heating. The purpose of this study is to develop an accurate and fast simulation method of this complex system which could be beneficial for system design, control and optimization for application. The present study upgraded the PV model by considering the variable resistance due to Peltier Effect in thermoelectric module. A new non-uniform time step model was proposed which can provide an improved and more efficient system simulation. The non-uniform time step solution of BIPVTE system was validated by comparing with both uniform time step solution and experimental data. The parametric studies on time step h and superposition number N under uniform time step solution, as well as two linear deviation coefficients d T and d G under non-uniform time step solution, were respectively analyzed. In uniform time step solution, the simulation time step h and parameter N should be properly chosen to balance simulation speed and accuracy. However, in non-uniform time step, numerical investigations demonstrated that simulation accuracy can be kept within an acceptable range even when linear deviation coefficients were large enough. The algorithm can be further accelerated by adopting Gauss-Berntsen-Espelid or Gauess-Kronrod rule in numerical integral calculation. The comparative and case study in this research has shown the validity and robustness of the proposed non-uniform time step model, which could be a useful tool for further work on BIPVTE as well as other building systems. [ABSTRACT FROM AUTHOR]