Search

This work presents a detailed case study on using Generative AI (GenAI) to develop AI surrogates for simulation models in fusion energy research. The scope includes the methodology, implementation, and results of using GenAI to assist in model development and optimization, comparing these results with previous manually developed models.

Wind power prediction is essential for ensuring the stability and efficient operation of modern power systems, particularly as renewable energy integration continues to expand. This paper presents a comprehensive review of machine learning techniques applied to wind power prediction, emphasizing their advantages over traditional physical and statistical models. Machine learning methods, especially deep learning approaches such as Convolutional Neural Networks (CNNs), Long Short-Term Memory Networks (LSTMs), and ensemble learning techniques like XGBoost, excel in addressing the nonlinearity and complexity of wind power data. The review also explores critical aspects such as data preprocessing, feature selection strategies, and model optimization techniques, which significantly enhance prediction accuracy and robustness. Challenges such as data acquisition difficulties, complex terrain influences, and sensor quality issues are examined in depth, with proposed solutions discussed. Additionally, the paper highlights future research directions, including the potential of multi-model fusion, emerging deep learning technologies like Transformers, and the integration of smart sensors and IoT technologies to develop intelligent, automated, and reliable prediction systems. By addressing existing challenges and leveraging advanced machine learning techniques, this work provides valuable insights into the current state of wind power prediction research and offers strategic guidance for enhancing the applicability and reliability of prediction models in practical scenarios. [ABSTRACT FROM AUTHOR]

Model optimization is a problem of great concern and challenge for developing an image classification model. In image classification, selecting the appropriate hyperparameters can substantially boost the model's ability to learn intricate patterns and features from complex image data. Hyperparameter optimization helps to prevent overfitting by finding the right balance between complexity and generalization of a model. The ensemble genetic algorithm and convolutional neural network (EGACNN) are proposed to enhance image classification by fine-tuning hyperparameters. The convolutional neural network (CNN) model is combined with a genetic algorithm GA) using stacking based on the Modified National Institute of Standards and Technology (MNIST) dataset to enhance efficiency and prediction rate on image classification. The GA optimizes the number of layers, kernel size, learning rates, dropout rates, and batch sizes of the CNN model to improve the accuracy and performance of the model. The objective of this research is to improve the CNN-based image classification system by utilizing the advantages of ensemble learning and GA. The highest accuracy is obtained using the proposed EGACNN model which is 99.91% and the ensemble CNN and spiking neural network (CSNN) model shows an accuracy of 99.68%. The ensemble approaches like EGACNN and CSNN tends to be more effective as compared to CNN, RNN, AlexNet, ResNet, and VGG models. The hyperparameter optimization of deep learning classification models reduces human efforts and produces better prediction results. Performance comparison with existing approaches also shows the superior performance of the proposed model. [ABSTRACT FROM AUTHOR]

Copyright of China Rural Water & Hydropower is the property of China Rural Water & Hydropower Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The existing detection algorithms are unable to achieve a suitable balance between detection accuracy and inference speed. As the accuracy of the algorithm increases, its complexity also rises, resulting in a decrease in detection speed, which undermines its practicality. This issue is particularly evident in the context of surface defect detection in industrial parts, where low contrast, small target features, difficult feature extraction, and low real-time detection efficiency are prominent challenges. This study proposes a novel method for steel defect detection based on the YOLO v8 algorithm, which improves detection accuracy while maintaining low computational complexity. Firstly, the global background and edge information are adaptively extracted via the MSA-SPPF module in order to obtain a more comprehensive feature representation. Furthermore, the anti-interference ability of the model is enhanced through the deformability of attention and the large convolution kernel characteristics. Secondly, the design of Dynamic Conv and C2f-OREPA enables the model to efficiently reduce the demand for computational resources while maintaining high performance. It is further proposed that the RepHead detection head approximates the multi-branch structure of the original training by a single convolution operation. This approach not only enriches the feature representation but also maintains an efficient inference process. The effectiveness of the improved MRP-YOLO algorithm is verified using the NEU-DET industrial surface defect dataset. The experimental results demonstrate that the mAP of the MRP-YOLO algorithm reaches 75.6%, which is 2.2% higher than that of the YOLOv8n algorithm, while the FLOPs are only 2.3 G higher. It indicates that the detection accuracy is significantly improved with a limited increase in computational complexity. [ABSTRACT FROM AUTHOR]

This paper explores the application of the YOLOv5s algorithm integrated with the DeepSORT tracking detection algorithm in vehicle target detection, leveraging its advantages in data processing, loss function, network structure, and training strategy. Regarding detection frame regression, adopting Focal-EIOU can improve vehicle detection accuracy by precisely measuring overlap and better handle complex scenarios, enhancing the overall performance. The CoordConv convolution layer with more spatial position information is employed to enhance the original network structure's convolution layer and improve vehicle positioning accuracy. The principle and effectiveness of the Shuffle Attention mechanism are analyzed and added to the YOLOv5s network structure to enhance training, improve detection accuracy and running speed. And the DeepSORT tracking detection algorithm is designed to achieve high-speed operation and high-accuracy matching in target tracking, enabling efficient and reliable tracking of objects. Simultaneously, the network structure is optimized to enhance algorithmic speed and performance. To meet the requirements of vehicle detection in practical transportation systems, real-world vehicle images are collected as a dataset for model training to achieve accurate vehicle detection. The results show that the accuracy rate P of the improved YOLOv5s algorithm is increased by 0.484%, and mAP_0.5:0.95 reaches 92.221%, with an increase of 1.747%. [ABSTRACT FROM AUTHOR]

Considering that most pavement anomaly detection algorithms are difficult to play a stable role in data related to different distributions of pavement anomalies, this paper proposes a pavement anomaly detection algorithm based on multi-scale fusion time series information. The algorithm iteratively extracts autoregressive latent variables at different time series scales to quantify dynamic regularity information. Then iterative feature set is embedded into the pictographic subsequence learning module to rapidly extract most representative subsequence features through deep learning. The static statistical evaluation value is retained as one of the specific data to assist in identifying the sample, and the two angle data complement each other to maximize the feature information entropy. Using the SVM classifier with the global sequence alignment algorithm as the main kernel function, the data classification module detects anomalies, so that the algorithm can mine valuable feature attributes from multiple angles as much as possible while stably exerting its classification function. The experimental results show that the multi-scale fusion features capture more complete information about the temporal trajectory than the single latent variable feature information, suggesting that the proposed algorithm exhibits better comprehensive performance and generalization ability than unimproved machine learning algorithms and typical deep learning classifiers.

Constructing a macromolecular model of coal is essential for conducting in-depth investigations into a myriad of coal-related physical processes and chemical reactions. These processes include the adsorption behavior of coalbed methane (CBM), the pyrolysis mechanisms of coal, and alterations in the microscopic wettability of coal dust. Such research has profound implications for the advancement of CBM mining and the coal chemical industry. Coal is a complex polymer. The cornerstone of constructing the coal macromolecular structure model lies in developing the optimization mechanism and validation methods, which collectively form the foundation for this molecular model. This paper highlights the critical significance and merits of employing molecular simulation techniques for the construction, optimization, and validation of coal macromolecular structure models. These techniques encompass molecular mechanics, molecular dynamics, and quantum mechanics. It also analyzes recent developments in classical coal macromolecular structure models and explores the potential applications of molecular simulation in this domain.

With the deep extension of coal mine mining, the influence of ground on occurrence state and regularity of coal seam gas is more and more obvious. In order to better study the kinetic characteristics of gas adsorption in the loaded coal, isothermal adsorption experiments were carried out in coal samples under different confining pressures and axial pressures. The quasi-first-order, quasi-second-order, Elovich and Banghamadsorption kinetic models were used to fit and analyze the experimental results, and the adsorption kinetic characteristics and influence mechanism of loaded coal were discussed. It provides a theoretical basis for further exploration of gas occurrence state and law in deep coal seam. Results indicate that as the adsorption progresses, the adsorption capacity increases and the trend gradually slows down and the rate of change of gas adsorption capacity is gradually reduced. The larger the applied load is, the smaller the equilibrium adsorption capacity of gas is. Among the four adsorption kinetic models, the Bangham kinetic model can better describe the dynamic characteristics of coal gas adsorption under the influence of confining pressure, and the quasi-first-order adsorption kinetic model can better describe the dynamic characteristics of coal gas adsorption under the influence of axial pressure. The equilibrium adsorption capacity parameter has a negative power function relationship with confining pressure and axial pressure. The adsorption rate parameters are negatively correlated with the applied load.

The article is dedicated to the analysis and modeling of the operation modes of low-flow wells with extended horizontal wellbore (HW) and multi-stage hydraulic fracturing (MHF) during artificial lift with an electric submersible pump (ESP). A comprehensive analysis of the operation modes of Bazhenov formation wells with HW and MHF, working with ESP, has been conducted. Complications in the operation of ESP in such wells, related to proppant carryover, constant productivity decline, and gas factor growth, are shown, and operational data of various types of ESP in wells with complex construction are provided. One of the main parts of the manuscript is the new methodology for creating a model of multiphase transient flow in a well with ESP operating in periodic short-term activation (PSA) mode using specialized software. Various approaches for numerical modeling of transient multiphase flows in a well with ESP and a bottomhole separator were used to accurately describe the physical processes. Difficulties that may arise during well operation were demonstrated through the analysis of field data and numerical modeling results. The work includes modeling of complications in HW due to proppant carryover from MHF fractures, resulting in flow blockage in the well. The article demonstrates the process of tuning transient operation modes of a well with ESP in PSA mode using specialized software based on real field data (multiphase flowmeter, pressure gauges, etc.), taking into account the well construction, ESP characteristics, additional equipment (valves, separator, etc.), and reservoir parameters. The model was tuned to actual data, and possible ways of optimizing the operation mode were shown. In addition, the work provides an analysis of the sensitivity of fluid properties and well parameters to non-stationary flow in the well.

Abstract With the rapid increase of users over social media, cyberbullying, and hate speech problems have arisen over the past years. Automatic hate speech detection (HSD) from text is an emerging research problem in natural language processing (NLP). Researchers developed various approaches to solve the automatic hate speech detection problem using different corpora in various languages, however, research on the Urdu language is rather scarce. This study aims to address the HSD task on Twitter using Roman Urdu text. The contribution of this research is the development of a hybrid model for Roman Urdu HSD, which has not been previously explored. The novel hybrid model integrates deep learning (DL) and transformer models for automatic feature extraction, combined with machine learning algorithms (MLAs) for classification. To further enhance model performance, we employ several hyperparameter optimization (HPO) techniques, including Grid Search (GS), Randomized Search (RS), and Bayesian Optimization with Gaussian Processes (BOGP). Evaluation is carried out on two publicly available benchmarks Roman Urdu corpora comprising HS-RU-20 corpus and RUHSOLD hate speech corpus. Results demonstrate that the Multilingual BERT (MBERT) feature learner, paired with a Support Vector Machine (SVM) classifier and optimized using RS, achieves state-of-the-art performance. On the HS-RU-20 corpus, this model attained an accuracy of 0.93 and an F1 score of 0.95 for the Neutral-Hostile classification task, and an accuracy of 0.89 with an F1 score of 0.88 for the Hate Speech-Offensive task. On the RUHSOLD corpus, the same model achieved an accuracy of 0.95 and an F1 score of 0.94 for the Coarse-grained task, alongside an accuracy of 0.87 and an F1 score of 0.84 for the Fine-grained task. These results demonstrate the effectiveness of our hybrid approach for Roman Urdu hate speech detection.

Abstract With the accelerated advancements in artificial intelligence and the increasing emphasis on sustainable supply chain management, the integration of multimodal artificial intelligence (AI) into green supply chains has emerged as a critical research frontier. This study delves into the synergistic potential and challenges of combining multimodal AI, which leverages diverse data types such as text, images, and numerical data, to enhance decision-making processes in green supply chains. Through the meticulous design of a data strategy and model framework, this research establishes a sophisticated and efficient data processing and model training pipeline. The experimental results reveal that the comprehensive analysis and fusion of multimodal data significantly improve the prediction accuracy of key supply chain metrics, with observed increases in accuracy and recall rates by 12.4% and 9.8%, respectively. Additionally, the model's limitations are critically assessed, and targeted improvement strategies are proposed. The practical implications of this study are profound, offering actionable insights for the application of multimodal AI in real-world energy sector scenarios. The findings underscore the potential of this technology to optimize operations, reduce environmental impact, and drive sustainable growth in the energy industry.

The speed of offsite consequence analysis is highly important due to the extensive calculations required to handle all the scenarios for a single-unit or multi-unit Level 3 PSA (probabilistic safety assessment). To perform an offsite consequence analysis as part of Level 3 PSA, various input parameters are considered, amongst which certain parameters, such as plume segments, spatial grids, and particle size distributions, have flexible input formats. This study describes the development of an effective optimization method to reduce the analysis time as much as possible while maintaining the accuracy of the offsite consequence analysis results. The effect of plume segmentation on offsite consequence analysis was investigated by observing deviations in analysis results and changes in the required analysis times following changes in plume release. Then a plume segmentation optimization method based on the cumulative release fraction slope was developed to intensively analyze the sections with rapid release and to simplify the analysis for the sections with nonsignificant release. As a result of applying this method, the analysis time was reduced by about 54.5 % compared to the base case, while the resulting health effects showed very small deviations of 0.03 % and 1.77 % for early fatality risk and cancer fatality risk, respectively.

Abstract This study explores the prediction of mechanical characteristics of linear polyethylene based on oven residence time, employing various regression models and hyper-parameter tuning through the Whale Optimization Algorithm. The dataset comprises one input variable (oven residence time) and three output parameters (Tensile Strength, Impact Strength, and Flexure Strength). The models investigated include Multilayer Perceptron, K-Nearest Neighbors, Support Vector Regression, Polynomial Regression, and Theil–Sen Regression. The results showcased distinct performances across the models for each output parameter. The Polynomial Regression (WOA-PR) method has been identified as the most suitable option for predicting Tensile Strength due to its ability to achieve the lowest errors in terms of Mean Absolute Error, Root Mean Square Error, and Average Absolute Relative Deviation. K-Nearest Neighbors (WOA-KNN) outperforms other models in predicting Impact Strength due to its superior accuracy and reliability. Additionally, Support Vector Regression (WOA-SVR) emerges as the best model for predicting Flexure Strength, showcasing notable performance in minimizing prediction errors. These findings underscore the significance of model selection and optimization techniques in accurately predicting the mechanical properties of polymers, paving the way for enhanced manufacturing processes and material design.

The rapid evolution of deep learning has led to significant achievements in computer vision, primarily driven by complex convolutional neural networks (CNNs). However, the increasing depth and parameter count of these networks often result in overfitting and elevated computational demands. Knowledge distillation (KD) has emerged as a promising technique to address these issues by transferring knowledge from a large, well-trained teacher model to a more compact student model. This paper introduces a novel knowledge distillation method that simplifies the distillation process and narrows the performance gap between teacher and student models without relying on intricate knowledge representations. Our approach leverages a unique teacher network architecture designed to enhance the efficiency and effectiveness of knowledge transfer. Additionally, we introduce a streamlined teacher network architecture that transfers knowledge effectively through a simplified distillation process, enabling the student model to achieve high accuracy with reduced computational demands. Comprehensive experiments conducted on the CIFAR-10 dataset demonstrate that our proposed model achieves superior performance compared to traditional KD methods and established architectures such as ResNet and VGG networks. The proposed method not only maintains high accuracy but also significantly reduces training and validation losses. Key findings highlight the optimal hyperparameter settings (temperature T = 15.0 and smoothing factor α = 0.7), which yield the highest validation accuracy and lowest loss values. This research contributes to the theoretical and practical advancements in knowledge distillation, providing a robust framework for future applications and research in neural network compression and optimization. The simplicity and efficiency of our approach pave the way for more accessible and scalable solutions in deep learning model deployment. [ABSTRACT FROM AUTHOR]

Copyright of China Rural Water & Hydropower is the property of China Rural Water & Hydropower Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

To address the issues of low processing efficiency, poor forming accuracy, and internal damage in glass material processing, this study proposes the use of flame jet forming. However, the mechanism of flame jet processing requires further elucidation. This research investigates the relationship between the indentation morphology on the glass surface and the inlet velocity of the flame jet. A theoretical model was established through mathematical analysis to reflect this relationship. The model's accuracy was validated using numerical simulation methods. By comparing experimental data with theoretical model results, surface tension was incorporated, and the model was iteratively optimized using MATLAB R2024a. The final optimized model demonstrated an absolute error range of 0.009 to 0.069 mm. This study confirms the feasibility of flame jet processing and enriches the understanding of its mechanism, providing a novel, efficient, and precise method for processing glass materials. [ABSTRACT FROM AUTHOR]

Copyright of Coal Geology & Exploration is the property of Xian Research Institute of China Coal Research Institute and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the increasing amount of devices connected to the huge net known as the internet, it is not surprising the corresponding growth of cyber attacks. The era of the Internet of Things (IoT) has proved to be an environment in which malicious activities have been able to proliferate with great comfort. Due to this kind of threat, having powerful machine learning-based classifiers is almost a critical need nowadays. Besides, to properly face new and previously unseen attacks unsupervised learners need to be involved. An unsupervised network threat detector is provided in this work. The proposed approach is based on three steps. The main one is a novel anomaly score, which relies on modeling the tails of the empirical distributions and on an interpretation of the well-known Bayes theorem. A step for an unsupervised feature selection and another one for data reduction are also considered, both aimed at gaining robustness of the procedure. All three steps are in a completely unsupervised way, facilitating an ad hoc scenario deploying. The whole method showed good performance reaching 98.44 % and 98.14 % in the F1-score over different datasets. The obtained results are competitive with other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Yoga has become an essential part of modern life, and hence, there has been a tremendous demand for self-training yoga platforms for trainer-less yoga practice. Robust and efficient recognition of yoga poses in video stream is the first requirement of such systems. However, the existing techniques for yoga pose recognition are compute-intensive and may fail in complex real-world conditions. These issues pose serious limitations on their practical applicability. To this end, this paper examines state-of-the-art deep learning techniques to implement a robust and compute-efficient system for yoga pose recognition in real-time on a resource-constrained embedded platform. The first technique uses a hybrid CNN & LSTM model, while the other three (3DCNN Model1, 3DCNN Model2, and 3DCNN Model3) employ the Sports1M pre-trained 3DCNN model named C3D. We assessed the performance of the designed architectures on a publicly available yoga pose database by applying four well-known metrics, namely recognition accuracy, precision, recall, and F1-score. On three database splits, the designed hybrid CNN & LSTM, 3DCNN Model1, 3DCNN Model2, and 3DCNN Model3 achieved mean recognition accuracy of 98.80%, 99.07% 98.19%, and 98.43%, respectively. Also, on one of the splits, the best-performing model achieved the highest recognition accuracy of 99.65% and, thus, surpassed the baseline accuracy of 99.38%. Also, the optimal model runs at a frame rate of 31 FPS on an Nvidia GPU-enabled desktop, much better than the previous best of 3 FPS. Finally, to evaluate the model's efficiency on embedded systems, we optimized it using TensorRT SDK and deployed it on an Nvidia Xavier embedded platform. The optimized model runs at 8 FPS on the resource-constrained embedded platform, demonstrating its suitability for real-world applications. A working demo of the developed system is available at https://youtu.be/at1GJ8Nxx38, and the source codes are available at https://github.com/sumeetssaurav/Yoga-Pose-Classification. [ABSTRACT FROM AUTHOR]

Image style transfer is a technique that applies artistic styles to natural images while maintaining their original content structures. This technology has a broad spectrum of applications in the stylization of images, comics, and videos, and has emerged as a significant area of research within the field of artificial intelligence. This paper seeks to advance research on deep learning-based image style transfer by summarizing and discussing the principal methods and illustrative outcomes within this domain. It commences with a review of traditional image style transfer algorithms, analyzing their strengths and limitations. The paper then delves into deep learning-based style transfer algorithms, classifying them into image optimization-based and model optimization-based approaches based on the efficiency of transfer. Through an analysis of the current state of image style transfer research, the paper outlines the developmental trajectory and recent milestones in the field. It evaluates the merits and demerits of various methods, addressing current issues such as suboptimal image style transfer quality, model specificity, and the need for enhanced semantic depth. Via comparative analysis, the paper proposes enhancement strategies and suggests future research directions. [ABSTRACT FROM AUTHOR]

With the accelerated advancements in artificial intelligence and the increasing emphasis on sustainable supply chain management, the integration of multimodal artificial intelligence (AI) into green supply chains has emerged as a critical research frontier. This study delves into the synergistic potential and challenges of combining multimodal AI, which leverages diverse data types such as text, images, and numerical data, to enhance decision-making processes in green supply chains. Through the meticulous design of a data strategy and model framework, this research establishes a sophisticated and efficient data processing and model training pipeline. The experimental results reveal that the comprehensive analysis and fusion of multimodal data significantly improve the prediction accuracy of key supply chain metrics, with observed increases in accuracy and recall rates by 12.4% and 9.8%, respectively. Additionally, the model's limitations are critically assessed, and targeted improvement strategies are proposed. The practical implications of this study are profound, offering actionable insights for the application of multimodal AI in real-world energy sector scenarios. The findings underscore the potential of this technology to optimize operations, reduce environmental impact, and drive sustainable growth in the energy industry. [ABSTRACT FROM AUTHOR]

In the dynamic operation of a boom sprayer, the boom oscillation will cause the detection value of the boom height to fluctuate greatly, resulting in failures of the control system. Based on the previously developed static boom height detection model for the entire wheat growth cycle, this study aimed to optimize the model to reduce the impact of boom oscillation on the accuracy of boom height detection. Three ultrasonic sensors were installed on each section boom of a three-section boom sprayer, and dynamic boom height detection tests were conducted at vehicle speeds of 4 to 8 km/h across six growth stages of winter wheat in Beijing, a total detection area within a single fixed operational row of approximately 14 ha. The test results showed that as vehicle speed increased, boom oscillations intensified across all sections. By setting the boom oscillation correction parameters, the detecting value of each section of boom height is corrected. The results show that the fluctuation and deviation degree of the boom height-detecting value are obviously reduced, and the correction effect is obvious. Further analysis of the detecting value of the boom height after the correction shows that the previously established detection model still maintains high detection accuracy under dynamic conditions; that is, the detection position of the ultrasonic sensor does not downward shift. This paper provides a low-cost technical method that can be directly applied to the dynamic detection of boom height. [ABSTRACT FROM AUTHOR]

The multi-scale object detection algorithm based on deep learning is a common method for safety monitoring in current power operation scenarios, which is of great significance to ensuring the safety of power operations. For certain power applications with high real-time requirements, the computational complexity of deep learning models may become a bottleneck. Deploying deep learning models requires high-performance hardware support, such as GPUs, which might not be easily achievable in some power field environments. According to the characteristics of common safety monitoring tasks in power operation scenarios, this paper proposes an automatic structured pruning method for multi-scale object detection algorithms. This method is designed for the safety monitoring of single and fixed-scale targets, effectively reducing the complexity of inference calculations and improving the frames per second (FPS) of real-time object detection without compromising accuracy. Furthermore, the proposed method can adaptively perform automatic structured pruning for targets of different scales. Experiments conducted using the YOLOv5 model demonstrate that the proposed method improves inference speed by approximately 20% for safety monitoring tasks without reducing detection accuracy. [ABSTRACT FROM AUTHOR]

Purpose: Forecasting outpatient volume during a significant security crisis can provide reasonable decision-making references for hospital managers to prevent sudden outbreaks and dispatch medical resources on time. Based on the background of standard hospital operation and Coronavirus disease (COVID-19) periods, this paper constructs a hybrid grey model to forecast the outpatient volume to provide foresight decision support for hospital decision-makers. Design/methodology/approach: This paper proposes an improved hybrid grey model for two stages. In the non-COVID-19 stage, the Aquila Optimizer (AO) is selected to optimize the modeling parameters. Fourier correction is applied to revise the stochastic disturbance. In the COVID-19 stage, this model adds the COVID-19 impact factor to improve the grey model forecasting results based on the dummy variables. The cycle of the dummy variables modifies the COVID-19 factor. Findings: This paper tests the hybrid grey model on a large Chinese hospital in Jiangsu. The fitting MAPE is 2.48%, and the RMSE is 16463.69 in the training group. The test MAPE is 1.91%, and the RMSE is 9354.93 in the test group. The results of both groups are better than those of the comparative models. Originality/value: The two-stage hybrid grey model can solve traditional hospitals' seasonal outpatient volume forecasting and provide future policy formulation references for sudden large-scale epidemics. [ABSTRACT FROM AUTHOR]

Introduction: Mechanical damage significantly reduces the market value of fruits, making the early detection of such damage a critical aspect of agricultural management. This study focuses on the early detection of mechanical damage in blueberries (variety: Sapphire) through a non-destructive method. Methods: The proposed method integrates hyperspectral image fusion with a multi-strategy improved support vector machine (SVM) model. Initially, spectral features and image features were extracted from the hyperspectral information using the successive projections algorithm (SPA) and Grey Level Co-occurrence Matrix (GLCM), respectively. Different models including SVM, RF (Random Forest), and PLS-DA (Partial Least Squares Discriminant Analysis) were developed based on the extracted features. To refine the SVM model, its hyperparameters were optimized using a multi-strategy improved Beluga Whale Optimization (BWO) algorithm. Results: The SVM model, upon optimization with the multi-strategy improved BWO algorithm, demonstrated superior performance, achieving the highest classification accuracy among the models tested. The optimized SVM model achieved a classification accuracy of 95.00% on the test set. Discussion: The integration of hyperspectral image information through feature fusion proved highly efficient for the early detection of bruising in blueberries. However, the effectiveness of this technology is contingent upon specific conditions in the detection environment, such as light intensity and temperature. The high accuracy of the optimized SVM model underscores its potential utility in post-harvest assessment of blueberries for early detection of bruising. Despite these promising results, further studies are needed to validate the model under varying environmental conditions and to explore its applicability to other fruit varieties. [ABSTRACT FROM AUTHOR]

In the study, the structural parameters of Zichang (ZC) coking coal from northern Shaanxi Province were examined. A theoretical calculation was employed to build a molecular structure model for ZC coal, as well as applying principles of quantum chemistry, the prediction of NMR spectrogram and density for the model was achieved, and the molecular chemical formula was C199H155O36N3. The molecular structure optimization and annealing kinetics calculations are based on molecular mechanics (MM) and molecular dynamics (MD). Subsequently, a representative simplified model was constructed using the aromatic structure as the fundamental unit. On this foundation, the electrostatic potential (ESP), atomic charge distribution, and energy level orbitals were analyzed for this simplified model. The outcomes of this research can serve as an essential guide for determining the reaction order of the active categories during the low-temperature oxidation process for ZC coking coal. [ABSTRACT FROM AUTHOR]

Phishing and spam have been a cybersecurity threat with the majority of breaches resulting from these types of social engineering attacks. Therefore, detection has been a long‐standing challenge for both academic and industry researcher. New and innovative approaches are required to keep up with the growing sophistication of threat actors. One such illumination which has vast potential are large language models (LLM). LLM emerged and already demonstrated their potential to transform society and provide new and innovative approaches to solve well‐established challenges. Phishing and spam have caused financial hardships and lost time and resources to email users all over the world and frequently serve as an entry point for ransomware threat actors. While detection approaches exist, especially heuristic‐based approaches, LLMs offer the potential to venture into a new unexplored area for understanding and solving this challenge. LLMs have rapidly altered the landscape from business, consumers, and throughout academia and demonstrate transformational potential to profoundly impact the society. Based on this, applying these new and innovative approaches to email detection is a rational next step in academic research. In this work, we present IPSDM, an improved phishing spam detection model based on fine‐tuning the BERT family of models to specifically detect phishing and spam emails. We demonstrate our fine‐tuned version, IPSDM, is able to better classify emails in both unbalanced and balanced datasets. Moreover, IPSDM consistently outperforms the baseline models in terms of classification accuracy, precision, recall, and F1‐score, while concurrently mitigating overfitting concerns. [ABSTRACT FROM AUTHOR]

In Reunion Island, most of household wastes (HW) are landfilled. Residual household wastes (RHW) represent 44.6% of HW produced and are totally landfilled. Indeed, significant environmental burdens are caused by current waste management system. Hence, a new system that will ensure the energy recovery of waste was implemented. With the aim of integrating gasification into the waste management system in Reunion Island, a study of gasification of RHW through modeling is suggested in this work. Indeed, a 3-step methodology was adopted, the first step is waste characterization, 3 mixtures of waste have been established based on local context and data from the last waste characterisation campaign in Reunion. The second step is the modeling of a stratified downdraft reactor using an optimized steady-state equilibrium-based and kinetic model. Optimization consists of using correction factors to modify the equilibrium constants of the methanation and water gas shift reactions. The models were validated against extensive experimental data from literature. In addition, the effect of equivalence ratio and moisture content was studied in the steady-state equilibrium model. Furthermore, a sensitivity analysis of the kinetic model was conducted by investigating the effect of solid residence time and reduction temperature on the composition of producer gas. [ABSTRACT FROM AUTHOR]