Pub Date : 2025-02-16DOI: 10.1016/j.compeleceng.2025.110180
Sandesh Bhaktha B , Satyam Sarma , M. Vamshik , Jeyaraj Pitchaimani , K.V. Gangadharan
This paper presents an innovative methodology for optimizing the design parameters of a 500 W low-speed outer rotor switched reluctance motor (OR-SRM) for an electric bicycle (E-bike) in accordance with a driving cycle. Design optimization of SRMs based on driving cycles has been minimally explored in the literature, with all existing research focusing exclusively on high-speed electric vehicle (EV) applications. These studies utilized computationally intensive dynamic current analysis methods to account for the significant dynamic effects incurred. Given the E-bike's low-speed characteristics, the present study mitigates the computational load of design optimization through static current analysis. A high torque density 8/18 OR-multi-teeth (MT) SRM topology has been proposed. The benefits of this topology, such as mass, cost, torque ripple reductions, and improved torque density, have been highlighted through a comparison with a conventional 6/10 OR-SRM topology. The reliability of the finite element analysis models used in this study is validated through experiments conducted on an 8/18 OR-MTSRM prototype. The multi-objective design optimization aims to maximize starting torque and minimize torque ripple and electromagnetic losses throughout the driving cycle. The efficacy of the optimization is confirmed by the enhancement in the performance parameters of the optimal design compared to the preliminary design.
{"title":"Driving cycle-centric design optimization and experimental validation of high torque density outer rotor 8/18 MTSRM for an E-Bike","authors":"Sandesh Bhaktha B , Satyam Sarma , M. Vamshik , Jeyaraj Pitchaimani , K.V. Gangadharan","doi":"10.1016/j.compeleceng.2025.110180","DOIUrl":"10.1016/j.compeleceng.2025.110180","url":null,"abstract":"<div><div>This paper presents an innovative methodology for optimizing the design parameters of a 500 W low-speed outer rotor switched reluctance motor (OR-SRM) for an electric bicycle (E-bike) in accordance with a driving cycle. Design optimization of SRMs based on driving cycles has been minimally explored in the literature, with all existing research focusing exclusively on high-speed electric vehicle (EV) applications. These studies utilized computationally intensive dynamic current analysis methods to account for the significant dynamic effects incurred. Given the E-bike's low-speed characteristics, the present study mitigates the computational load of design optimization through static current analysis. A high torque density 8/18 OR-multi-teeth (MT) SRM topology has been proposed. The benefits of this topology, such as mass, cost, torque ripple reductions, and improved torque density, have been highlighted through a comparison with a conventional 6/10 OR-SRM topology. The reliability of the finite element analysis models used in this study is validated through experiments conducted on an 8/18 OR-MTSRM prototype. The multi-objective design optimization aims to maximize starting torque and minimize torque ripple and electromagnetic losses throughout the driving cycle. The efficacy of the optimization is confirmed by the enhancement in the performance parameters of the optimal design compared to the preliminary design.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110180"},"PeriodicalIF":4.0,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420233","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To meet the efficient energy conversion in dual active bridge (DAB) converter-based microgrids, an optimum duty ratio generation is the foremost challenge for the researchers. Therefore, to optimize the current stress on DABs and minimize the conduction losses & reactive power, the current study proposes a novel minimum current point tracking (MCPT) assisted adaptive dual phase shift (DPS) controller for the microgrids. The proposed controller incorporates the simple structure and the conventional perturb and observe (P&O) algorithm to track the active power accurately and transmit it for the optimum phase shift generation. Due to the simple structure of the P&O algorithm, the introduced controller exhibits parameter independence and eliminates the requirement of convoluted system modeling. A comparative study of the introduced controller with its extended phase shift (EPS) counterpart is put through to show the promising performance of the proposed controller. Simulation studies are carried out in the MATLAB/Simulink environment, whereas, the experimental studies are carried away on the DAB prototype regulated through a microcontroller (TI-piccolo 280049C).
{"title":"An adaptive dual phase shift controller for the dual active bridge converter applied in active power transmission and control of microgrid","authors":"Anupam Kumar , Hiramani Shukla , Diwaker Pathak , Subhanarayan Sahoo , Shubhendra Pratap Singh","doi":"10.1016/j.compeleceng.2025.110183","DOIUrl":"10.1016/j.compeleceng.2025.110183","url":null,"abstract":"<div><div>To meet the efficient energy conversion in dual active bridge (DAB) converter-based microgrids, an optimum duty ratio generation is the foremost challenge for the researchers. Therefore, to optimize the current stress on DABs and minimize the conduction losses & reactive power, the current study proposes a novel minimum current point tracking (MCPT) assisted adaptive dual phase shift (DPS) controller for the microgrids. The proposed controller incorporates the simple structure and the conventional perturb and observe (P&O) algorithm to track the active power accurately and transmit it for the optimum phase shift generation. Due to the simple structure of the P&O algorithm, the introduced controller exhibits parameter independence and eliminates the requirement of convoluted system modeling. A comparative study of the introduced controller with its extended phase shift (EPS) counterpart is put through to show the promising performance of the proposed controller. Simulation studies are carried out in the MATLAB/Simulink environment, whereas, the experimental studies are carried away on the DAB prototype regulated through a microcontroller (TI-piccolo 280049C).</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110183"},"PeriodicalIF":4.0,"publicationDate":"2025-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143422309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1016/j.compeleceng.2025.110145
Gang Shen , Kongze Xiao , Jun Tu , Hua Shen , Mingwu Zhang
Relying on wireless communication technology, vehicular ad-hoc networks (VANETs) realizes vehicle-to-everything (V2X) communication. It can collect real-time data from vehicles and roads to help drivers deal with emergencies and reduce the risk of accidents. However, these multi-dimensional real-time data might reveal the sensitive information of vehicles. Meanwhile, unreliable real-time data can also cause serious accidents. To combat these challenges, we propose a privacy-preserving and robust aggregation scheme for multi-dimensional data in VANETs. Specifically, we utilize the Chinese remainder theorem (CRT) to convert multi-dimensional real-time data into a large integer to alleviate the burden of its processing, and an EC-ElGamal homomorphic encryption with dual trapdoor decryption mechanism is used to realize the decryption and data query function both vehicles and traffic management center (TMC) to meet the needs of different entities. In addition, Schnorr signature is applied to prevent data from being tampered with or forged, and a credibility mechanism is proposed to guarantee the validity of the data. Security analysis shows that our scheme can ensure the security of vehicle identity and data privacy, and has the ability to identify data availability. Finally, we conduct a theoretical analysis of performance, and use the simulation framework to evaluate the communication overhead. Compared with other schemes, our scheme reduces computational cost and communication overhead by at most 99.6% and 88%, respectively.
{"title":"A privacy-preserving and robust aggregation scheme for multi-dimensional data in VANETs","authors":"Gang Shen , Kongze Xiao , Jun Tu , Hua Shen , Mingwu Zhang","doi":"10.1016/j.compeleceng.2025.110145","DOIUrl":"10.1016/j.compeleceng.2025.110145","url":null,"abstract":"<div><div>Relying on wireless communication technology, vehicular ad-hoc networks (VANETs) realizes vehicle-to-everything (V2X) communication. It can collect real-time data from vehicles and roads to help drivers deal with emergencies and reduce the risk of accidents. However, these multi-dimensional real-time data might reveal the sensitive information of vehicles. Meanwhile, unreliable real-time data can also cause serious accidents. To combat these challenges, we propose a privacy-preserving and robust aggregation scheme for multi-dimensional data in VANETs. Specifically, we utilize the Chinese remainder theorem (CRT) to convert multi-dimensional real-time data into a large integer to alleviate the burden of its processing, and an EC-ElGamal homomorphic encryption with dual trapdoor decryption mechanism is used to realize the decryption and data query function both vehicles and traffic management center (TMC) to meet the needs of different entities. In addition, Schnorr signature is applied to prevent data from being tampered with or forged, and a credibility mechanism is proposed to guarantee the validity of the data. Security analysis shows that our scheme can ensure the security of vehicle identity and data privacy, and has the ability to identify data availability. Finally, we conduct a theoretical analysis of performance, and use the simulation framework to evaluate the communication overhead. Compared with other schemes, our scheme reduces computational cost and communication overhead by at most 99.6% and 88%, respectively.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110145"},"PeriodicalIF":4.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420153","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1016/j.compeleceng.2025.110112
Sanfeng Zhang , Baiyu Tian , Yang Gao , Mengyao Dai , Wang Yang
Recent advancements in stylish content generation technologies have revolutionized image synthesis, enabling the creation of high-resolution images with precise control. Motivated by this progress, we present BirdsSong, a novel audio steganography framework. BirdsSong aims to improve steganalysis resistance, increase steganographic capacity, enhance audio quality, and diversify styles compared to traditional methods. The framework includes an audio style extractor to separate content and style, a generator to combine style and content vectors without damaging audio quality, and a message extractor to recover the secret message accurately. Experimental results demonstrate that BirdsSong outperforms baseline methods in steganographic capacity, audio quality, style diversity, and resistance to steganalysis tools.
{"title":"BirdsSong: A stylized generative audio steganography","authors":"Sanfeng Zhang , Baiyu Tian , Yang Gao , Mengyao Dai , Wang Yang","doi":"10.1016/j.compeleceng.2025.110112","DOIUrl":"10.1016/j.compeleceng.2025.110112","url":null,"abstract":"<div><div>Recent advancements in stylish content generation technologies have revolutionized image synthesis, enabling the creation of high-resolution images with precise control. Motivated by this progress, we present BirdsSong, a novel audio steganography framework. BirdsSong aims to improve steganalysis resistance, increase steganographic capacity, enhance audio quality, and diversify styles compared to traditional methods. The framework includes an audio style extractor to separate content and style, a generator to combine style and content vectors without damaging audio quality, and a message extractor to recover the secret message accurately. Experimental results demonstrate that BirdsSong outperforms baseline methods in steganographic capacity, audio quality, style diversity, and resistance to steganalysis tools.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110112"},"PeriodicalIF":4.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143420152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recommendation systems are decision-support systems used by e-commerce enterprises to evaluate customers’ preferences and suggest items based on their interests. Moreover, they also tackle the problem of information overload. Multi-criteria recommendation systems differ from standard approaches by using multiple-criterion ratings instead of single-criterion ratings while rating products. Multi-criteria recommendation systems has attracted significant attention in the field of recommendation systems due to the lower accuracy of single-criteria recommendation systems. Furthermore, deep learning models have demonstrated encouraging results in several domains including image processing, computer vision, pattern recognition, and natural language processing. This paper introduces a novel methodology leveraging deep autoencoders optimized using the Non-dominated sorting genetic algorithm (NSGA-II) a meta-heuristic optimization technique to enhance the accuracy of multi-criteria recommendation systems. In the first stage of the proposed method, NSGA-II is employed to optimize the weights of the deep autoencoder for enhancing the fine-tuning of the hyperparameters. Secondly, missing ratings and overall ratings are predicted using autoencoders for more precise top-N recommendations. The model is validated using two real-world multi-criteria datasets Yahoo! Movies and TripAdvisor. Experimental results demonstrate that the model achieves significant improvements in prediction accuracy, with a Mean Absolute Error (MAE) of 0.6012 and 0.7215, and Root Mean Squared Error (RMSE) of 0.6137 and 0.7362 on Yahoo! Movies and TripAdvisor datasets, respectively. These findings indicate that the model outperforms both single-criteria recommendation algorithms and other state-of-the-art multi-criteria recommendation models in accuracy.
{"title":"NSGA-II optimized deep autoencoders for enhanced multi-criteria recommendation system","authors":"Ishwari Singh Rajput , Anand Shanker Tewari , Arvind Kumar Tiwari","doi":"10.1016/j.compeleceng.2025.110159","DOIUrl":"10.1016/j.compeleceng.2025.110159","url":null,"abstract":"<div><div>Recommendation systems are decision-support systems used by e-commerce enterprises to evaluate customers’ preferences and suggest items based on their interests. Moreover, they also tackle the problem of information overload. Multi-criteria recommendation systems differ from standard approaches by using multiple-criterion ratings instead of single-criterion ratings while rating products. Multi-criteria recommendation systems has attracted significant attention in the field of recommendation systems due to the lower accuracy of single-criteria recommendation systems. Furthermore, deep learning models have demonstrated encouraging results in several domains including image processing, computer vision, pattern recognition, and natural language processing. This paper introduces a novel methodology leveraging deep autoencoders optimized using the Non-dominated sorting genetic algorithm (NSGA-II) a meta-heuristic optimization technique to enhance the accuracy of multi-criteria recommendation systems. In the first stage of the proposed method, NSGA-II is employed to optimize the weights of the deep autoencoder for enhancing the fine-tuning of the hyperparameters. Secondly, missing ratings and overall ratings are predicted using autoencoders for more precise top-N recommendations. The model is validated using two real-world multi-criteria datasets Yahoo! Movies and TripAdvisor. Experimental results demonstrate that the model achieves significant improvements in prediction accuracy, with a Mean Absolute Error (MAE) of 0.6012 and 0.7215, and Root Mean Squared Error (RMSE) of 0.6137 and 0.7362 on Yahoo! Movies and TripAdvisor datasets, respectively. These findings indicate that the model outperforms both single-criteria recommendation algorithms and other state-of-the-art multi-criteria recommendation models in accuracy.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110159"},"PeriodicalIF":4.0,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143395687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The failure of the most potent medicines to eradicate superbugs underscores the urgent need to develop new antimicrobial drugs. Antibacterial peptides (ABPs) are oligopeptides present in all multicellular organisms and serve as the first line of defense against pathogens. ABPs provide several benefits over conventional antibiotics; therefore, they have recently gained significant attention as an alternative. Finding ABPs in the laboratory is expensive and time-consuming. Therefore, wet-lab researchers use in-silico tools to discover ABPs from natural sources. The existing tools available for this purpose suffer from the limitation of being black boxes. In the present work, we developed XAI-INVENT, an explainable artificial intelligence-based framework for the rapid discovery of novel antibiotics. For building XAI-INVENT, first, the probability scores of deep learning models are fused, and then the fused scores are utilized with local interpretable model-agnostic explanations (LIME) for determining the critical amino acids. The value of performance metrics, namely Accuracy, Sensitivity, Precision, F1-Score, Specificity, and Matthews correlation coefficient obtained by the proposed framework for test data is 96 %, 96 %, 97 %, 96 %, 97 %, and 92 %, respectively. To help wet-lab researchers, XAI-INVENT is deployed as a web server at https://xai-invent.anvil.app/.
{"title":"XAI-INVENT: An explainable artificial intelligence based framework for rapid discovery of novel antibiotics","authors":"Ritesh Sharma , Sameer Shrivastava , Sanjay Kumar Singh , Abhinav Kumar , Amit Kumar Singh , Sonal Saxena","doi":"10.1016/j.compeleceng.2025.110098","DOIUrl":"10.1016/j.compeleceng.2025.110098","url":null,"abstract":"<div><div>The failure of the most potent medicines to eradicate superbugs underscores the urgent need to develop new antimicrobial drugs. Antibacterial peptides (ABPs) are oligopeptides present in all multicellular organisms and serve as the first line of defense against pathogens. ABPs provide several benefits over conventional antibiotics; therefore, they have recently gained significant attention as an alternative. Finding ABPs in the laboratory is expensive and time-consuming. Therefore, wet-lab researchers use <em>in-silico</em> tools to discover ABPs from natural sources. The existing tools available for this purpose suffer from the limitation of being black boxes. In the present work, we developed XAI-INVENT, an explainable artificial intelligence-based framework for the rapid discovery of novel antibiotics. For building XAI-INVENT, first, the probability scores of deep learning models are fused, and then the fused scores are utilized with local interpretable model-agnostic explanations (LIME) for determining the critical amino acids. The value of performance metrics, namely Accuracy, Sensitivity, Precision, F1-Score, Specificity, and Matthews correlation coefficient obtained by the proposed framework for test data is <span><math><mo>≈</mo></math></span> 96 %, 96 %, 97 %, 96 %, 97 %, and 92 %, respectively. To help wet-lab researchers, XAI-INVENT is deployed as a web server at <span><span>https://xai-invent.anvil.app/</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110098"},"PeriodicalIF":4.0,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143387042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-12DOI: 10.1016/j.compeleceng.2025.110140
Hamideh Zebardast, Mahmoud Farhang, Abbas Sheikhi
In multiple-input multiple-output (MIMO) radars, achieving accurate target detection and interference mitigation necessitates transmitting sequences with minimal sidelobes at the matched filters output. In this paper, we propose an approach to design phase-only (unimodular) sequences that minimize the peak sidelobe level after matched filtering (PSLF). With the PSLF as the key figure of merit, we cast the design problem into an -norm minimization framework. To tackle the inherent unimodularity constraint, we introduce a product manifold within the unimodular sequence space and develop efficient manifold-based optimization algorithms. Numerical results show that, compared to existing methods, the proposed algorithms achieve significantly better performance in terms of both PSLF value and computational efficiency.
{"title":"A new approach to low peak sidelobe level signal design in MIMO radars","authors":"Hamideh Zebardast, Mahmoud Farhang, Abbas Sheikhi","doi":"10.1016/j.compeleceng.2025.110140","DOIUrl":"10.1016/j.compeleceng.2025.110140","url":null,"abstract":"<div><div>In multiple-input multiple-output (MIMO) radars, achieving accurate target detection and interference mitigation necessitates transmitting sequences with minimal sidelobes at the matched filters output. In this paper, we propose an approach to design phase-only (unimodular) sequences that minimize the peak sidelobe level after matched filtering (PSLF). With the PSLF as the key figure of merit, we cast the design problem into an <span><math><msub><mrow><mi>l</mi></mrow><mrow><mi>p</mi></mrow></msub></math></span>-norm minimization framework. To tackle the inherent unimodularity constraint, we introduce a product manifold within the unimodular sequence space and develop efficient manifold-based optimization algorithms. Numerical results show that, compared to existing methods, the proposed algorithms achieve significantly better performance in terms of both PSLF value and computational efficiency.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110140"},"PeriodicalIF":4.0,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143387041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-12DOI: 10.1016/j.compeleceng.2025.110147
Weiwei Bai , Guoqiang Zheng , Weibing Xia , Yu Mu , Yujun Xue
Aiming at the problem of system performance degradation caused by dynamic spectrum access in underlay mode within cognitive radio networks, we propose a multi-agent deep reinforcement learning-based joint channel selection and power control (MA-JCSPC) method. This method formulates the spectrum access problem in underlay mode as an optimization problem of joint channel selection and power control, and transforms this optimization problem into a multi-agent Markov decision process. By designing a multi-agent deep reinforcement learning framework with centralized training and decentralized execution, the channel selection and power control strategies for secondary users are optimized. In this process, a nonlinear reward function is designed by introducing a penalty term, and a novel initial action selection strategy based on a action guidance term is employed to solve the sparse rewards and ineffective exploration problems. The simulation results demonstrate that the MA-JCSPC method surpasses the compared methods in convergence, resource allocation rationality, and throughput. Compared to the centralized deep reinforcement learning (C-DRL) method, the proposed method achieves an average improvement of 6.7% and 9.1% in the sum throughput of secondary users, respectively, under variations in the throughput requirements of the primary user and the number of secondary users.
{"title":"Multi-agent deep reinforcement learning-based joint channel selection and power control method","authors":"Weiwei Bai , Guoqiang Zheng , Weibing Xia , Yu Mu , Yujun Xue","doi":"10.1016/j.compeleceng.2025.110147","DOIUrl":"10.1016/j.compeleceng.2025.110147","url":null,"abstract":"<div><div>Aiming at the problem of system performance degradation caused by dynamic spectrum access in underlay mode within cognitive radio networks, we propose a multi-agent deep reinforcement learning-based joint channel selection and power control (MA-JCSPC) method. This method formulates the spectrum access problem in underlay mode as an optimization problem of joint channel selection and power control, and transforms this optimization problem into a multi-agent Markov decision process. By designing a multi-agent deep reinforcement learning framework with centralized training and decentralized execution, the channel selection and power control strategies for secondary users are optimized. In this process, a nonlinear reward function is designed by introducing a penalty term, and a novel initial action selection strategy based on a action guidance term is employed to solve the sparse rewards and ineffective exploration problems. The simulation results demonstrate that the MA-JCSPC method surpasses the compared methods in convergence, resource allocation rationality, and throughput. Compared to the centralized deep reinforcement learning (C-DRL) method, the proposed method achieves an average improvement of 6.7% and 9.1% in the sum throughput of secondary users, respectively, under variations in the throughput requirements of the primary user and the number of secondary users.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110147"},"PeriodicalIF":4.0,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143395454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-12DOI: 10.1016/j.compeleceng.2025.110122
Wenhui Zhu , Houjun Li , Xiande Bu , Lei Xu , Aerduoni Jiu , Chunxia Dou
This paper proposes a novel framework for ultra-short-term distributed photovoltaic (PV) power prediction, aiming to improve prediction accuracy and reliability, ensuring the safe, stable, and economically efficient operation of active distribution networks. This framework uniquely integrates data augmentation, clustering, and quantum machine learning (QML). Firstly, considering the problem of insufficient data under extreme weather fluctuation patterns, the Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) incorporating bidirectional Long Short-Term Memory (BiLSTM) layers is adopted for data expansion. Introducing BiLSTM network layers enhances its ability to capture long-term sequence dependencies. Secondly, a two-stage clustering method is specifically designed to classify weather fluctuation patterns accurately. On this basis, a hybrid quantum–classical prediction model is constructed by combining Variational Quantum Circuits (VQC) with Long Short-Term Memory (LSTM) networks to compensate for the shortcomings of traditional methods in feature mining. In addition, this article introduces a new evaluation metric: the Improved Weighted Mean Absolute Percentage Error (WMAPE-), which is used to measure model performance more comprehensively. The comparative experiments indicate that the proposed model outperforms BiLSTM, LSTM, DLinear, Gated Recurrent Unit (GRU), Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU), and Temporal Convolutional Network (TCN) models in terms of prediction accuracy, convergence speed, stability, and generalization capability. Under different weather fluctuation patterns, the average values of the proposed model are 0.998, 0.993, and 0.984, respectively. This study provides a new reference direction for accurate prediction of distributed PV power, which is of great significance for optimizing grid integration and energy management in renewable energy systems.
{"title":"HWDQT: A hybrid quantum machine learning method for ultra-short-term distributed photovoltaic power prediction","authors":"Wenhui Zhu , Houjun Li , Xiande Bu , Lei Xu , Aerduoni Jiu , Chunxia Dou","doi":"10.1016/j.compeleceng.2025.110122","DOIUrl":"10.1016/j.compeleceng.2025.110122","url":null,"abstract":"<div><div>This paper proposes a novel framework for ultra-short-term distributed photovoltaic (PV) power prediction, aiming to improve prediction accuracy and reliability, ensuring the safe, stable, and economically efficient operation of active distribution networks. This framework uniquely integrates data augmentation, clustering, and quantum machine learning (QML). Firstly, considering the problem of insufficient data under extreme weather fluctuation patterns, the Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) incorporating bidirectional Long Short-Term Memory (BiLSTM) layers is adopted for data expansion. Introducing BiLSTM network layers enhances its ability to capture long-term sequence dependencies. Secondly, a two-stage clustering method is specifically designed to classify weather fluctuation patterns accurately. On this basis, a hybrid quantum–classical prediction model is constructed by combining Variational Quantum Circuits (VQC) with Long Short-Term Memory (LSTM) networks to compensate for the shortcomings of traditional methods in feature mining. In addition, this article introduces a new evaluation metric: the Improved Weighted Mean Absolute Percentage Error (WMAPE-<span><math><mi>β</mi></math></span>), which is used to measure model performance more comprehensively. The comparative experiments indicate that the proposed model outperforms BiLSTM, LSTM, DLinear, Gated Recurrent Unit (GRU), Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU), and Temporal Convolutional Network (TCN) models in terms of prediction accuracy, convergence speed, stability, and generalization capability. Under different weather fluctuation patterns, the average <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> values of the proposed model are 0.998, 0.993, and 0.984, respectively. This study provides a new reference direction for accurate prediction of distributed PV power, which is of great significance for optimizing grid integration and energy management in renewable energy systems.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110122"},"PeriodicalIF":4.0,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143387040","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-11DOI: 10.1016/j.compeleceng.2025.110132
John Enriquez-Loja, Bryan Castillo-Pérez, Xavier Serrano-Guerrero, Antonio Barragán-Escandón
This study presents a comprehensive methodology to objectively evaluate various clustering techniques applied to electrical demand profiles (EDPs). The effectiveness of Self-Organizing Maps (SOM), Fuzzy C-Means (FCM), and Hierarchical Clustering (HC) is analyzed, revealing that these methods achieve anomalous value percentages below 17.1%. The proposed approach includes a statistical framework based on confidence intervals to classify data as typical or atypical, thereby facilitating the selection of the most appropriate clustering technique based on the characteristics of the dataset. To evaluate the methodology, an analysis of probability distributions is used, comparing it with three internal validation techniques through the implementation of specific criteria. These metrics provide insight into the dispersion and distribution of the EDPs, allowing for a robust evaluation of how variations in data impact clustering outcomes. The results indicate that the SOM, FCM and HC techniques exhibit strong adaptability to different patterns of variability, making them suitable for diverse applications in energy management. This research contributes valuable tools for optimizing the classification of EDPs, enhancing the understanding of consumption behaviors in the electricity sector.
{"title":"Performance evaluation method for different clustering techniques","authors":"John Enriquez-Loja, Bryan Castillo-Pérez, Xavier Serrano-Guerrero, Antonio Barragán-Escandón","doi":"10.1016/j.compeleceng.2025.110132","DOIUrl":"10.1016/j.compeleceng.2025.110132","url":null,"abstract":"<div><div>This study presents a comprehensive methodology to objectively evaluate various clustering techniques applied to electrical demand profiles (EDPs). The effectiveness of Self-Organizing Maps (SOM), Fuzzy C-Means (FCM), and Hierarchical Clustering (HC) is analyzed, revealing that these methods achieve anomalous value percentages below 17.1%. The proposed approach includes a statistical framework based on confidence intervals to classify data as typical or atypical, thereby facilitating the selection of the most appropriate clustering technique based on the characteristics of the dataset. To evaluate the methodology, an analysis of probability distributions is used, comparing it with three internal validation techniques through the implementation of specific criteria. These metrics provide insight into the dispersion and distribution of the EDPs, allowing for a robust evaluation of how variations in data impact clustering outcomes. The results indicate that the SOM, FCM and HC techniques exhibit strong adaptability to different patterns of variability, making them suitable for diverse applications in energy management. This research contributes valuable tools for optimizing the classification of EDPs, enhancing the understanding of consumption behaviors in the electricity sector.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"123 ","pages":"Article 110132"},"PeriodicalIF":4.0,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143378650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号