Pub Date : 2025-12-01Epub Date: 2025-09-03DOI: 10.1016/j.egyai.2025.100612
Zhuo Liu , Bumin Meng , Rui Pan , Juan Zhou
Retired batteries for secondary use offer significant economic benefits and environmental value. Accurate sorting of retired batteries with diverse characteristics can further enhance their application efficiency. However, in practical sorting processes, the presence of redundant features, noise interference, and distribution discrepancies in the data severely limits the accuracy of sorting outcomes. To address these challenges, this paper proposes an enhanced retired battery sorting strategy that incorporates feature selection and a clustering algorithm, aiming to optimize the sorting process from the perspective of feature data. To address feature redundancy and high dimensionality issues, this paper proposes an entropy screening method. The Local Outlier Factor algorithm is used to remove anomalous samples. Subsequently, an ensemble clustering approach is developed based on K-means, Density-Based Spatial Clustering of Applications with Noise, Gaussian Mixture Model, and Spectral clustering, to handle diverse data distributions. The proposed method is validated on 100 retired batteries as well as the large-scale dataset. Additionally, its strong sorting capability and engineering applicability are further demonstrated through carefully designed aging-controlled experiments.
{"title":"An enhanced sorting framework for retired batteries based on multi-dimensional features and an integrated clustering approach","authors":"Zhuo Liu , Bumin Meng , Rui Pan , Juan Zhou","doi":"10.1016/j.egyai.2025.100612","DOIUrl":"10.1016/j.egyai.2025.100612","url":null,"abstract":"<div><div>Retired batteries for secondary use offer significant economic benefits and environmental value. Accurate sorting of retired batteries with diverse characteristics can further enhance their application efficiency. However, in practical sorting processes, the presence of redundant features, noise interference, and distribution discrepancies in the data severely limits the accuracy of sorting outcomes. To address these challenges, this paper proposes an enhanced retired battery sorting strategy that incorporates feature selection and a clustering algorithm, aiming to optimize the sorting process from the perspective of feature data. To address feature redundancy and high dimensionality issues, this paper proposes an entropy screening method. The Local Outlier Factor algorithm is used to remove anomalous samples. Subsequently, an ensemble clustering approach is developed based on K-means, Density-Based Spatial Clustering of Applications with Noise, Gaussian Mixture Model, and Spectral clustering, to handle diverse data distributions. The proposed method is validated on 100 retired batteries as well as the large-scale dataset. Additionally, its strong sorting capability and engineering applicability are further demonstrated through carefully designed aging-controlled experiments.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100612"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145010678","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-10-16DOI: 10.1016/j.egyai.2025.100630
Abdullah Shoukat , Zhongyong Liu , Yousif Yahia Ahmed Abuker , Jingguo Li , Lei Mao
Proton exchange membrane fuel cell (PEMFC) faults, especially dehydration and flooding, cause distinct changes in electrochemical behavior. Consequently, real-time monitoring is essential for early and accurate diagnosis. However, acquiring real-world fault data is challenging, and the rarity of such faults results in severe class imbalance. This imbalance limits the performance, reliability, and practical applicability of conventional diagnostic methods. To address these limitations, this study proposes a unified diagnostic framework that integrates multi-sine AC voltage response, boundary-aware resampling, and attention-guided generative modeling. The key innovations of the proposed approach include: (1) Enhanced fault separability through the first application of multi-sine AC voltage response under data imbalance, enabling real-time extraction of critical electrochemical spectral features for early-stage diagnosis; (2) Improved data balance and clearer class boundaries using synthetic minority oversampling with Tomek links, which oversamples minority classes and removes borderline samples; (3) Realistic minority class synthesis using a dual attention Wasserstein generative adversarial networks, where channel attention focuses on diagnostically relevant spectral features and temporal attention models the dynamic evolution of PEMFC electrochemical behavior, ensuring high-quality, diagnostically informative synthetic fault data. The integrated framework achieves 99.67 % overall diagnostic accuracy and, under an extreme 1:200 class imbalance, outperforms state-of-the-art methods by 14 %. This approach enables rapid, data-efficient PEMFC fault diagnosis, strengthening fault management and advancing the performance of energy systems.
{"title":"Dual attention-enhanced data augmentation for diagnosing water management faults in proton exchange membrane fuel cells using imbalanced multi-sine AC data","authors":"Abdullah Shoukat , Zhongyong Liu , Yousif Yahia Ahmed Abuker , Jingguo Li , Lei Mao","doi":"10.1016/j.egyai.2025.100630","DOIUrl":"10.1016/j.egyai.2025.100630","url":null,"abstract":"<div><div>Proton exchange membrane fuel cell (PEMFC) faults, especially dehydration and flooding, cause distinct changes in electrochemical behavior. Consequently, real-time monitoring is essential for early and accurate diagnosis. However, acquiring real-world fault data is challenging, and the rarity of such faults results in severe class imbalance. This imbalance limits the performance, reliability, and practical applicability of conventional diagnostic methods. To address these limitations, this study proposes a unified diagnostic framework that integrates multi-sine AC voltage response, boundary-aware resampling, and attention-guided generative modeling. The key innovations of the proposed approach include: (1) Enhanced fault separability through the first application of multi-sine AC voltage response under data imbalance, enabling real-time extraction of critical electrochemical spectral features for early-stage diagnosis; (2) Improved data balance and clearer class boundaries using synthetic minority oversampling with Tomek links, which oversamples minority classes and removes borderline samples; (3) Realistic minority class synthesis using a dual attention Wasserstein generative adversarial networks, where channel attention focuses on diagnostically relevant spectral features and temporal attention models the dynamic evolution of PEMFC electrochemical behavior, ensuring high-quality, diagnostically informative synthetic fault data. The integrated framework achieves 99.67 % overall diagnostic accuracy and, under an extreme 1:200 class imbalance, outperforms state-of-the-art methods by 14 %. This approach enables rapid, data-efficient PEMFC fault diagnosis, strengthening fault management and advancing the performance of energy systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100630"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145362267","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-08-21DOI: 10.1016/j.egyai.2025.100575
Nejira Hadzalic , Jacob Hamar , Marco Fischer , Simon Erhard , Jan Philipp Schmidt
Battery electric vehicles are exposed to highly diverse operating conditions and driving behaviors that strongly influence degradation pathways, yet these real-world complexities are only partially captured in laboratory aging tests. This study investigates a semi-supervised learning approach for robust estimation of battery state of health, defined as the ratio of remaining to nominal capacity. The method integrates a multi-view co-training algorithm with a rule-based pseudo labeling mechanism and is developed and validated using field data from 3000 BMW i3 vehicles with battery capacity of 60 Ah, collected since 2013 across 34 countries. The available data comprises standardized full charge capacity measurements, which serve as ground truth labels. The proposed training and validation pipeline is designed to address challenges inherent in real-world data generation and is particularly advantageous during early deployment of new battery technologies, when labeled data is scarce. By incrementally incorporating newly available labeled data into both evaluation and retraining, the model adapts to heterogeneous aging patterns observed in the field. Comparative analysis demonstrates that, relative to a supervised benchmark, the proposed method reduces estimation error by 28 % under limited-label conditions and by 6 % under optimally labeled scenarios, highlighting its robustness for field applications.
{"title":"Semi-supervised battery state of health estimation for field applications","authors":"Nejira Hadzalic , Jacob Hamar , Marco Fischer , Simon Erhard , Jan Philipp Schmidt","doi":"10.1016/j.egyai.2025.100575","DOIUrl":"10.1016/j.egyai.2025.100575","url":null,"abstract":"<div><div>Battery electric vehicles are exposed to highly diverse operating conditions and driving behaviors that strongly influence degradation pathways, yet these real-world complexities are only partially captured in laboratory aging tests. This study investigates a semi-supervised learning approach for robust estimation of battery state of health, defined as the ratio of remaining to nominal capacity. The method integrates a multi-view co-training algorithm with a rule-based pseudo labeling mechanism and is developed and validated using field data from 3000 BMW i3 vehicles with battery capacity of 60<!--> <!-->Ah, collected since 2013 across 34 countries. The available data comprises standardized full charge capacity measurements, which serve as ground truth labels. The proposed training and validation pipeline is designed to address challenges inherent in real-world data generation and is particularly advantageous during early deployment of new battery technologies, when labeled data is scarce. By incrementally incorporating newly available labeled data into both evaluation and retraining, the model adapts to heterogeneous aging patterns observed in the field. Comparative analysis demonstrates that, relative to a supervised benchmark, the proposed method reduces estimation error by 28<!--> <!-->% under limited-label conditions and by 6<!--> <!-->% under optimally labeled scenarios, highlighting its robustness for field applications.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100575"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144908147","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-08-27DOI: 10.1016/j.egyai.2025.100599
Yuemeng Zhang , Longqin Guo , Zeqian Chen , Hongtao Yan , Le Liang , Chunjing Lin
Accurate prediction of electric vehicle (EV) charging duration is critical for improving user satisfaction and enabling efficient real-time charging management. This paper proposes a dynamic charging duration prediction framework for EVs, composed of four coordinated modules: data preprocessing, charging pattern classification, static prediction, and dynamic bias correction. First, raw charging data collected from the Battery Management System (BMS) is cleaned and normalized to address missing and abnormal values. An enhanced convolutional autoencoder (EV-CAE) is then employed to extract multi-scale temporal features, while K-Means clustering is used to identify representative charging behavior patterns. Based on the classified patterns, the static prediction module estimates the current charging duration by leveraging historical data and pattern labels. To enhance adaptability under dynamic conditions, a bias correction mechanism is designed, integrating linear, logarithmic, proportional, and deep learning-based strategies to adjust the prediction results in real time. Experimental results on real-world EV datasets demonstrate that the proposed framework significantly improves prediction accuracy. In particular, the dynamic correction module increases the coefficient of determination (R²) from 0.948 to 0.960, while maintaining robust performance under fluctuating charging behavior and low-temperature conditions. These results validate the practical applicability and engineering potential of the proposed method for real-time charging duration estimation in intelligent EV charging systems.
{"title":"A novel framework for vehicle charging pattern recognition and charging duration prediction based on EA-CAE and K-means clustering","authors":"Yuemeng Zhang , Longqin Guo , Zeqian Chen , Hongtao Yan , Le Liang , Chunjing Lin","doi":"10.1016/j.egyai.2025.100599","DOIUrl":"10.1016/j.egyai.2025.100599","url":null,"abstract":"<div><div>Accurate prediction of electric vehicle (EV) charging duration is critical for improving user satisfaction and enabling efficient real-time charging management. This paper proposes a dynamic charging duration prediction framework for EVs, composed of four coordinated modules: data preprocessing, charging pattern classification, static prediction, and dynamic bias correction. First, raw charging data collected from the Battery Management System (BMS) is cleaned and normalized to address missing and abnormal values. An enhanced convolutional autoencoder (EV-CAE) is then employed to extract multi-scale temporal features, while K-Means clustering is used to identify representative charging behavior patterns. Based on the classified patterns, the static prediction module estimates the current charging duration by leveraging historical data and pattern labels. To enhance adaptability under dynamic conditions, a bias correction mechanism is designed, integrating linear, logarithmic, proportional, and deep learning-based strategies to adjust the prediction results in real time. Experimental results on real-world EV datasets demonstrate that the proposed framework significantly improves prediction accuracy. In particular, the dynamic correction module increases the coefficient of determination (R²) from 0.948 to 0.960, while maintaining robust performance under fluctuating charging behavior and low-temperature conditions. These results validate the practical applicability and engineering potential of the proposed method for real-time charging duration estimation in intelligent EV charging systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100599"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145045868","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Developing accurate and computationally efficient dynamic models for packed-bed latent-heat storages (PBLHS) is crucial for reliably predicting their performance across different operating scenarios and enabling their use in planning and real-time control. In this study, a novel neural-accelerated numerical model for PBLHS is proposed by coupling a neural network (NN) into a coarsely discretized equations of the Continuous-solid Phase (CP) model. The embedded NN predicts the surface temperature of the phase change material (PCM) given the fluid temperature and enthalpy of the PCM as inputs, which the CP model fails to capture. This allows the neural-accelerated model to replicate the accuracy of a high-fidelity and computationally expensive model namely Concentric Dispersion (CD) model. An innovative data generation process to generate training data for NN involving both CD and CP model is proposed. Two versions of neural-accelerated model are proposed, one with conventional NN and another using NN with a custom activation function. Both versions demonstrate an excellent accuracy, achieving MSE as low as 0.117 °C, values closer to 0.995 and error percentage below 0.394 compared to the highly accurate CD model. As for computational efficiency, the proposed models achieved 342 times and 764 times acceleration respectively. The gain in more acceleration for the later version of the proposed model is achieved through the use of a compact architecture that benefits from the custom activation function, while also enhancing model explainability. These results highlight the model’s suitability for scenarios demanding both high accuracy and computational efficiency.
{"title":"Neural-accelerated numerical model for packed bed latent heat storage system","authors":"Dessie Tadele Embiale , Shri Balaji Padmanabhan , Mohamed Tahar Mabrouk , Stéphane Grieu , Bruno Lacarrière","doi":"10.1016/j.egyai.2025.100602","DOIUrl":"10.1016/j.egyai.2025.100602","url":null,"abstract":"<div><div>Developing accurate and computationally efficient dynamic models for packed-bed latent-heat storages (PBLHS) is crucial for reliably predicting their performance across different operating scenarios and enabling their use in planning and real-time control. In this study, a novel neural-accelerated numerical model for PBLHS is proposed by coupling a neural network (NN) into a coarsely discretized equations of the Continuous-solid Phase (CP) model. The embedded NN predicts the surface temperature of the phase change material (PCM) given the fluid temperature and enthalpy of the PCM as inputs, which the CP model fails to capture. This allows the neural-accelerated model to replicate the accuracy of a high-fidelity and computationally expensive model namely Concentric Dispersion (CD) model. An innovative data generation process to generate training data for NN involving both CD and CP model is proposed. Two versions of neural-accelerated model are proposed, one with conventional NN and another using NN with a custom activation function. Both versions demonstrate an excellent accuracy, achieving MSE as low as 0.117 °C, <span><math><msup><mrow><mi>R</mi></mrow><mn>2</mn></msup></math></span> values closer to 0.995 and error percentage below 0.394<span><math><mo>%</mo></math></span> compared to the highly accurate CD model. As for computational efficiency, the proposed models achieved 342 times and 764 times acceleration respectively. The gain in more acceleration for the later version of the proposed model is achieved through the use of a compact architecture that benefits from the custom activation function, while also enhancing model explainability. These results highlight the model’s suitability for scenarios demanding both high accuracy and computational efficiency.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100602"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144921292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-08DOI: 10.1016/j.egyai.2025.100616
Yuanjing Zhuo, Huan Long, Zhi Wu, Wei Gu
Distributed photovoltaic (PV) power forecasting in newly installed systems faces challenges due to inherent stochastic volatilities and limited historical data. This paper proposes a lightweight feature transfer learning (LFTL) method that enables rapid and accurate forecasting of new distributed PVs. Firstly, the raw fluctuating PV data are preprocessed through decomposition to separate low- and high-frequency components. These components are then multi-scale segmented to capture diverse temporal characteristics. Following feature compression and LSTM temporal modeling, the informative features from the source domain enable lightweight transfer. For the target domain, a channel-independent encoder is designed to prevent negative interactions between heterogeneous frequencies. The frequency-fused segment-independent decoder equipped with positional embeddings enables local temporal analysis and reduces error accumulation of multi-step forecasts. LFTL trains with a joint training strategy to avoid negative transfer caused by domain disparity. LFTL consistently outperforms state-of-the-art time-series forecast models while maintaining a relatively low computational overhead based on real-world distributed PV data.
{"title":"LFTL: Lightweight feature transfer learning with channel-independent LSTM for distributed PV forecasting","authors":"Yuanjing Zhuo, Huan Long, Zhi Wu, Wei Gu","doi":"10.1016/j.egyai.2025.100616","DOIUrl":"10.1016/j.egyai.2025.100616","url":null,"abstract":"<div><div>Distributed photovoltaic (PV) power forecasting in newly installed systems faces challenges due to inherent stochastic volatilities and limited historical data. This paper proposes a lightweight feature transfer learning (LFTL) method that enables rapid and accurate forecasting of new distributed PVs. Firstly, the raw fluctuating PV data are preprocessed through decomposition to separate low- and high-frequency components. These components are then multi-scale segmented to capture diverse temporal characteristics. Following feature compression and LSTM temporal modeling, the informative features from the source domain enable lightweight transfer. For the target domain, a channel-independent encoder is designed to prevent negative interactions between heterogeneous frequencies. The frequency-fused segment-independent decoder equipped with positional embeddings enables local temporal analysis and reduces error accumulation of multi-step forecasts. LFTL trains with a joint training strategy to avoid negative transfer caused by domain disparity. LFTL consistently outperforms state-of-the-art time-series forecast models while maintaining a relatively low computational overhead based on real-world distributed PV data.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100616"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-16DOI: 10.1016/j.egyai.2025.100605
Paolo De Angelis , Giulio Barletta , Giovanni Trezza , Pietro Asinari , Eliodoro Chiavazzo
Artificial Intelligence (AI) in materials science is driving significant advancements in the discovery of advanced materials for energy applications. The recent GNoME protocol identifies over 380,000 novel stable crystals. From this, we identify over 38,500 materials with potential as energy materials forming the core of the Energy-GNoME database. Our unique combination of Machine Learning (ML) and Deep Learning (DL) tools mitigates cross-domain data bias using feature spaces, thus identifying potential candidates for thermoelectric materials, novel battery cathodes, and novel perovskites. First, classifiers with both structural and compositional features detect domains of applicability, where we expect enhanced reliability of regressors. Here, regressors are trained to predict key materials properties, like thermoelectric figure of merit (), band gap (), and cathode voltage (). This method significantly narrows the pool of potential candidates, serving as an efficient guide for experimental and computational chemistry investigations and accelerating the discovery of materials suited for electricity generation, energy storage and conversion.
{"title":"Energy-GNoME: A living database of selected materials for energy applications","authors":"Paolo De Angelis , Giulio Barletta , Giovanni Trezza , Pietro Asinari , Eliodoro Chiavazzo","doi":"10.1016/j.egyai.2025.100605","DOIUrl":"10.1016/j.egyai.2025.100605","url":null,"abstract":"<div><div>Artificial Intelligence (AI) in materials science is driving significant advancements in the discovery of advanced materials for energy applications. The recent GNoME protocol identifies over 380,000 novel stable crystals. From this, we identify over 38,500 materials with potential as energy materials forming the core of the Energy-GNoME database. Our unique combination of Machine Learning (ML) and Deep Learning (DL) tools mitigates cross-domain data bias using feature spaces, thus identifying potential candidates for thermoelectric materials, novel battery cathodes, and novel perovskites. First, classifiers with both structural and compositional features detect domains of applicability, where we expect enhanced reliability of regressors. Here, regressors are trained to predict key materials properties, like thermoelectric figure of merit (<span><math><mrow><mi>z</mi><mi>T</mi></mrow></math></span>), band gap (<span><math><msub><mrow><mi>E</mi></mrow><mrow><mi>g</mi></mrow></msub></math></span>), and cathode voltage (<span><math><mrow><mi>Δ</mi><msub><mrow><mi>V</mi></mrow><mrow><mi>c</mi></mrow></msub></mrow></math></span>). This method significantly narrows the pool of potential candidates, serving as an efficient guide for experimental and computational chemistry investigations and accelerating the discovery of materials suited for electricity generation, energy storage and conversion.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100605"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145096080","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-08-28DOI: 10.1016/j.egyai.2025.100564
Adrien Pavão , Antoine Marot , Jules Sintes , Viktor Eriksson Möllerstedt , Laure Crochepierre , Karim Chaouache , Benjamin Donnot , Van Tuan Dang , Isabelle Guyon
Achieving carbon neutrality by 2050 will require power-grid operators to absorb unprecedented volumes of variable solar and wind generation while maintaining reliability. To tackle this systems-level bottleneck, Réseau de Transport d’Électricité (RTE) and the research community launched Learn To Run A Power Network (L2RPN), a crowd-sourced competition aiming to accelerate the integration of intermittent renewables into power-grid operations. L2RPN is based on 16 years of weekly scenarios (832 in total) on a 118-node grid under realistic constraints, and casts real-time grid operation as a Markov-Decision-Process. The six participating teams tackled the challenge by developing autonomous agents with various strategies blending heuristics, optimization, data scaling, supervised learning, and reinforcement learning. We provide a detailed overview of all six participants’ performance under the competition’s demanding design. In addition, we present an in-depth analysis of the winning solution – made publicly available – which achieves consistent decision making across scenarios, executes real-time multimodal actions in under five seconds, and performs efficient topology control via action-space reduction and a neural policy that predicts useful grid actions with over 80% accuracy. In parallel, we trained a neural alert module on 315,000 samples derived from top agents, achieving 93.9% recall in flagging dangerous states and allowing agents to predict future failure. Finally, this work not only demonstrates AI’s promise and current limits in real-time grid management but also lays a transparent foundation for more robust, trustworthy systems in the energy transition.
{"title":"AI challenge for safe and low carbon power grid operation","authors":"Adrien Pavão , Antoine Marot , Jules Sintes , Viktor Eriksson Möllerstedt , Laure Crochepierre , Karim Chaouache , Benjamin Donnot , Van Tuan Dang , Isabelle Guyon","doi":"10.1016/j.egyai.2025.100564","DOIUrl":"10.1016/j.egyai.2025.100564","url":null,"abstract":"<div><div>Achieving carbon neutrality by 2050 will require power-grid operators to absorb unprecedented volumes of variable solar and wind generation while maintaining reliability. To tackle this systems-level bottleneck, Réseau de Transport d’Électricité (RTE) and the research community launched Learn To Run A Power Network (L2RPN), a crowd-sourced competition aiming to accelerate the integration of intermittent renewables into power-grid operations. L2RPN is based on 16 years of weekly scenarios (832 in total) on a 118-node grid under realistic constraints, and casts real-time grid operation as a Markov-Decision-Process. The six participating teams tackled the challenge by developing autonomous agents with various strategies blending heuristics, optimization, data scaling, supervised learning, and reinforcement learning. We provide a detailed overview of all six participants’ performance under the competition’s demanding design. In addition, we present an in-depth analysis of the winning solution – made publicly available – which achieves consistent decision making across scenarios, executes real-time multimodal actions in under five seconds, and performs efficient topology control via action-space reduction and a neural policy that predicts useful grid actions with over 80% accuracy. In parallel, we trained a neural alert module on 315,000 samples derived from top agents, achieving 93.9% recall in flagging dangerous states and allowing agents to predict future failure. Finally, this work not only demonstrates AI’s promise and current limits in real-time grid management but also lays a transparent foundation for more robust, trustworthy systems in the energy transition.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100564"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144988855","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-13DOI: 10.1016/j.egyai.2025.100619
Yeong Geon Son, Sung-Yul Kim
The reliable and coordinated operation of energy systems is becoming increasingly important as renewable energy penetration grows and electricity and gas infrastructures become more interconnected. This study addresses the challenge of aligning multiple stakeholders’ objectives in integrated electricity and gas distribution systems by proposing a sequential constrained optimization method. The method solves the multi-objective optimization problem by sequentially prioritizing each entity’s objective while incorporating others as adaptive-weighted sub-objectives and constraints. This process ensures that all entities participate in a fair and balanced decision-making procedure, ultimately converging to a consensus-based solution. The algorithm is validated using IEEE 33-bus and 118-bus test systems coupled with gas networks. Results show that the proposed method improves optimal resource allocation effectiveness by up to 3.66 compared to individual-objective or aggregated-objective benchmarks. Specifically, the method achieves performance improvements ranging from 0.02 pu to 1.7 pu across four distinct entities, highlighting its superiority in balancing conflicting operational goals. Moreover, the method demonstrates low computational delay and converges in fewer than 15 iterations for all tested cases. The algorithm adapts flexibly to different system configurations and maintains solution stability even under asymmetric stakeholder preferences. These findings indicate that the proposed sequential constrained optimization framework is a scalable and effective approach for equitable, multi-agent coordination in integrated multi-energy systems.
{"title":"Sequential constrained optimization for multi-entity operation of integrated electricity-gas distribution systems","authors":"Yeong Geon Son, Sung-Yul Kim","doi":"10.1016/j.egyai.2025.100619","DOIUrl":"10.1016/j.egyai.2025.100619","url":null,"abstract":"<div><div>The reliable and coordinated operation of energy systems is becoming increasingly important as renewable energy penetration grows and electricity and gas infrastructures become more interconnected. This study addresses the challenge of aligning multiple stakeholders’ objectives in integrated electricity and gas distribution systems by proposing a sequential constrained optimization method. The method solves the multi-objective optimization problem by sequentially prioritizing each entity’s objective while incorporating others as adaptive-weighted sub-objectives and constraints. This process ensures that all entities participate in a fair and balanced decision-making procedure, ultimately converging to a consensus-based solution. The algorithm is validated using IEEE 33-bus and 118-bus test systems coupled with gas networks. Results show that the proposed method improves optimal resource allocation effectiveness by up to 3.66 compared to individual-objective or aggregated-objective benchmarks. Specifically, the method achieves performance improvements ranging from 0.02 pu to 1.7 pu across four distinct entities, highlighting its superiority in balancing conflicting operational goals. Moreover, the method demonstrates low computational delay and converges in fewer than 15 iterations for all tested cases. The algorithm adapts flexibly to different system configurations and maintains solution stability even under asymmetric stakeholder preferences. These findings indicate that the proposed sequential constrained optimization framework is a scalable and effective approach for equitable, multi-agent coordination in integrated multi-energy systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100619"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118544","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-02DOI: 10.1016/j.egyai.2025.100606
Qing Wang , Lizhen Wu , Qiang Zheng , Liang An
As a key pathway for green hydrogen production, water electrolysis is expected to play a central role in the future energy landscape. However, its large-scale deployment is hindered by challenges related to cost, performance, and durability. The emergence of artificial intelligence (AI) has transformed this field by offering powerful and efficient tools for the design and optimization of electrocatalysts. This review outlines an AI-driven multiscale design framework, highlighting its role at the microscopic scale for identifying atomic-level active sites and key descriptors, at the mesoscopic scale for structural and morphological characterization, and at the macroscopic scale for multi-objective optimization and intelligent control. This multiscale framework demonstrates the potential of AI to accelerate the development of next-generation electrocatalysts. In addition, the integration of generative AI and automated experimental techniques is highlighted as promising strategies to further enhance electrocatalyst discovery and promote the practical implementation of water electrolysis technologies.
{"title":"Opportunities and perspectives of artificial intelligence in electrocatalysts design for water electrolysis","authors":"Qing Wang , Lizhen Wu , Qiang Zheng , Liang An","doi":"10.1016/j.egyai.2025.100606","DOIUrl":"10.1016/j.egyai.2025.100606","url":null,"abstract":"<div><div>As a key pathway for green hydrogen production, water electrolysis is expected to play a central role in the future energy landscape. However, its large-scale deployment is hindered by challenges related to cost, performance, and durability. The emergence of artificial intelligence (AI) has transformed this field by offering powerful and efficient tools for the design and optimization of electrocatalysts. This review outlines an AI-driven multiscale design framework, highlighting its role at the microscopic scale for identifying atomic-level active sites and key descriptors, at the mesoscopic scale for structural and morphological characterization, and at the macroscopic scale for multi-objective optimization and intelligent control. This multiscale framework demonstrates the potential of AI to accelerate the development of next-generation electrocatalysts. In addition, the integration of generative AI and automated experimental techniques is highlighted as promising strategies to further enhance electrocatalyst discovery and promote the practical implementation of water electrolysis technologies.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"22 ","pages":"Article 100606"},"PeriodicalIF":9.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145007632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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