Energy and AI最新文献

{"title":"Ignition delay prediction for fuels with diverse molecular structures using transfer learning-based neural networks","authors":"Mo Yang,&nbsp;Dezhi Zhou","doi":"10.1016/j.egyai.2024.100467","DOIUrl":"10.1016/j.egyai.2024.100467","url":null,"abstract":"<div><div>In this study, a transfer learning-based neural network approach to predict ignition delays for a variety of fuels is proposed to meet the demand for accurate combustion analysis. A comprehensive dataset of ignition delays was generated using a random sampling technique across different temperatures and pressures, focusing on hydrocarbon fuels with 1–4 carbon atoms. Two machine learning models, an artificial neural network and a graph convolutional network, are trained on this dataset, and their prediction performance was evaluated. A transfer learning framework was subsequently developed, enabling the models trained on smaller molecules (1–3 carbon atoms) to predict ignition delays for larger molecules (4 carbon atoms) with minimal additional data. The proposed framework demonstrated reliable and high prediction accuracy, achieving a high level of reliability for fuels with limited experimental measurements. This approach offers significant potential to streamline the prediction of ignition delays for novel fuels, reducing the dependence on resource-intensive experiments and complex simulations while contributing to the advancement of clean and efficient energy technologies.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100467"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"RL-ADN: A high-performance Deep Reinforcement Learning environment for optimal Energy Storage Systems dispatch in active distribution networks","authors":"Shengren Hou ,&nbsp;Shuyi Gao ,&nbsp;Weijie Xia ,&nbsp;Edgar Mauricio Salazar Duque ,&nbsp;Peter Palensky ,&nbsp;Pedro P. Vergara","doi":"10.1016/j.egyai.2024.100457","DOIUrl":"10.1016/j.egyai.2024.100457","url":null,"abstract":"<div><div>Deep Reinforcement Learning (DRL) presents a promising avenue for optimizing Energy Storage Systems (ESSs) dispatch in distribution networks. This paper introduces RL-ADN, an innovative open-source library specifically designed for solving the optimal ESSs dispatch in active distribution networks. RL-ADN offers unparalleled flexibility in modeling distribution networks, and ESSs, accommodating a wide range of research goals. A standout feature of RL-ADN is its data augmentation module, based on Gaussian Mixture Model and Copula (GMC) functions, which elevates the performance ceiling of DRL agents, achieving an average performance improvement of 21.43%, 1.08%, 2.76%, by augmenting five-year, one-year and three-month data, respectively. Additionally, RL-ADN incorporates the Tensor Power Flow solver, significantly reducing the computational burden of power flow calculations during training without sacrificing accuracy, maintaining voltage magnitude with an average error not exceeding 0.0001%. The effectiveness of RL-ADN is demonstrated using distribution networks with size varying, showing marked performance improvements in the adaptability of DRL algorithms for ESS dispatch tasks. Furthermore, RL-ADN achieves a tenfold increase in computational efficiency during training, making it highly suitable for large-scale network applications. The library sets a new benchmark in DRL-based ESSs dispatch in distribution networks and it is poised to advance DRL applications in distribution network operations significantly. RL-ADN is available at: <span><span>https://github.com/ShengrenHou/RL-ADN</span><svg><path></path></svg></span> and <span><span>https://github.com/distributionnetworksTUDelft/RL-ADN</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100457"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DGL-STFA: Predicting lithium-ion battery health with dynamic graph learning and spatial–temporal fusion attention","authors":"Zheng Chen ,&nbsp;Quan Qian","doi":"10.1016/j.egyai.2024.100462","DOIUrl":"10.1016/j.egyai.2024.100462","url":null,"abstract":"<div><div>Accurately predicting the State of Health (SOH) of lithium-ion batteries is a critical challenge to ensure their reliability and safety in energy storage systems, such as electric vehicles and renewable energy grids. The intricate battery degradation process is influenced by evolving spatial and temporal interactions among health indicators. Existing methods often fail to capture the dynamic interactions between health indicators over time, resulting in limited predictive accuracy. To address these challenges, we propose a novel framework, Dynamic Graph Learning with Spatial–Temporal Fusion Attention (DGL-STFA), which transforms health indicator time-series data into time-evolving graph representations. The framework employs multi-scale convolutional neural networks to capture diverse temporal patterns, a self-attention mechanism to construct dynamic adjacency matrices that adapt over time, and a temporal attention mechanism to identify and prioritize key moments that influence battery degradation. This combination enables DGL-STFA to effectively model both dynamic spatial relationships and long-term temporal dependencies, enhancing SOH prediction accuracy. Extensive experiments were conducted on the NASA and CALCE battery datasets, comparing this framework with traditional time-series prediction methods and other graph-based prediction methods. The results demonstrate that our framework significantly improves prediction accuracy, with a mean absolute error more than 30% lower than other methods. Further analysis demonstrated the robustness of DGL-STFA across various battery life stages, including early, mid, and end-of-life phases. These results highlight the capability of DGL-STFA to accurately predict SOH, addressing critical challenges in advancing battery health monitoring for energy storage applications.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100462"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Probabilistic forecasting of renewable energy and electricity demand using Graph-based Denoising Diffusion Probabilistic Model","authors":"Amir Miraki ,&nbsp;Pekka Parviainen ,&nbsp;Reza Arghandeh","doi":"10.1016/j.egyai.2024.100459","DOIUrl":"10.1016/j.egyai.2024.100459","url":null,"abstract":"<div><div>Renewable energy production and the balance between production and demand have become increasingly crucial in modern power systems, necessitating accurate forecasting. Traditional deterministic methods fail to capture the inherent uncertainties associated with intermittent renewable sources and fluctuating demand patterns. This paper proposes a novel denoising diffusion method for multivariate time series probabilistic forecasting that explicitly models the interdependencies between variables through graph modeling. Our framework employs a parallel feature extraction module that simultaneously captures temporal dynamics and spatial correlations, enabling improved forecasting accuracy. Through extensive evaluation on two real-world datasets focused on renewable energy and electricity demand, we demonstrate that our approach achieves state-of-the-art performance in probabilistic energy time series forecasting tasks. By explicitly modeling variable interdependencies and incorporating temporal information, our method provides reliable probabilistic forecasts, crucial for effective decision-making and resource allocation in the energy sector. Extensive experiments validate that our proposed method reduces the Continuous Ranked Probability Score (CRPS) by 2.1%–70.9%, Mean Absolute Error (MAE) by 4.4%–52.2%, and Root Mean Squared Error (RMSE) by 7.9%–53.4% over existing methods on two real-world datasets.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100459"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155833","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostic machine learning models for thermophysical characteristics of nanodiamond-based nanolubricants for heat pump systems","authors":"Ammar M. Bahman ,&nbsp;Emil Pradeep ,&nbsp;Zafar Said ,&nbsp;Prabhakar Sharma","doi":"10.1016/j.egyai.2024.100453","DOIUrl":"10.1016/j.egyai.2024.100453","url":null,"abstract":"<div><div>Lubricants for compressor oil significantly enhance the energy efficiency and performance of heat pump (HP) systems. This study compares prognostic machine learning (ML) models designed to predict the thermal conductivity and viscosity of nanolubricants used in HP compressors. Nanodiamond (ND) nanoparticles were mixed in Polyolester (POE) oil at volume concentrations ranging from 0.05 to 0.5 vol.% and temperatures ranging from 10 to 100<!--> <!-->°C. The data collected from the experimental research were used to build prognostic models using modern supervised ML techniques, including Gaussian process regression (GPR) and boosted regression tree (BRT). The GPR model demonstrated superior performance compared to the BRT model, achieving coefficient of correlation (R) values of 0.9996 and 0.9991 for thermal conductivity and viscosity, respectively. The reliability of the GPR and BRT models was further validated through comprehensive validation, sensitivity analysis, and extrapolation assessment using both empirical and unseen dataset references from the literature. When validated against an empirical correlation, the ML models exhibited a mean absolute error (MAE) of 0.17% for thermal conductivity and below 8% for viscosity. Additionally, when the GPR-based model was extended up to 120<!--> <!-->°C, the parametric analysis confirmed the reliability and accuracy of thermal conductivity and viscosity within a relative error of 5%. Furthermore, in the extrapolation analysis, despite changes in oil grade and nanolubricant concentrations, the GPR-based model showed a maximum absolute error (AE) of 19% compared to non-trained experimental data. Overall, the developed ML models can aid in designing and optimizing ND/POE nanolubricants for HP applications, achieving desired performance parameters while remaining economically viable and reducing the need for time-consuming laboratory-based testing.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100453"},"PeriodicalIF":9.6,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143131781","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning for battery quality classification and lifetime prediction using formation data","authors":"Jiayu Zou ,&nbsp;Yingbo Gao ,&nbsp;Moritz H. Frieges ,&nbsp;Martin F. Börner ,&nbsp;Achim Kampker ,&nbsp;Weihan Li","doi":"10.1016/j.egyai.2024.100451","DOIUrl":"10.1016/j.egyai.2024.100451","url":null,"abstract":"<div><div>Accurate classification of battery quality and prediction of battery lifetime before leaving the factory would bring economic and safety benefits. Here, we propose a data-driven approach with machine learning to classify the battery quality and predict the battery lifetime before usage only using formation data. We extract three classes of features from the raw formation data, considering the statistical aspects, differential analysis, and electrochemical characteristics. The correlation between over 100 extracted features and the battery lifetime is analysed based on the ageing mechanisms. Machine learning models are developed to classify battery quality and predict battery lifetime by features with a high correlation with battery ageing. The validation results show that the quality classification model achieved accuracies of 89.74% and 89.47% for the batteries aged at 25°C and 45°C, respectively. Moreover, the lifetime prediction model is able to predict the battery end-of-life with mean percentage errors of 6.50% and 5.45% for the batteries aged at 25°C and 45°C, respectively. This work highlights the potential of battery formation data from production lines in quality classification and lifetime prediction.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100451"},"PeriodicalIF":9.6,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142745679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Neural network potential-based molecular investigation of thermal decomposition mechanisms of ethylene and ammonia","authors":"Zhihao Xing,&nbsp;Rodolfo S.M. Freitas,&nbsp;Xi Jiang","doi":"10.1016/j.egyai.2024.100454","DOIUrl":"10.1016/j.egyai.2024.100454","url":null,"abstract":"<div><div>This study developed neural network potentials (NNPs) specifically tailored for pure ethylene and ethylene-ammonia blended systems for the first time. The NNPs were trained on a dataset generated from density functional theory (DFT) calculations, combining the computational accuracy of DFT with a calculation speed comparable to reactive force field methods. The NNPs are employed in reactive molecular dynamics simulations to explore the thermal decomposition reaction mechanisms of ethylene and ammonia. The simulation results revealed that adding ammonia reduces the activation energy for ethylene decomposition, thereby accelerating ethylene consumption. Furthermore, the addition of ammonia uncovers a new reaction pathway for hydrogen radical consumption, which reduces the occurrence of H-abstraction reactions from ethylene by hydrogen radicals. The inhibition effect of ammonia addition on soot formation mainly acts in two aspects: on the one hand, ammonia decomposition products react with carbon-containing species, ultimately producing C₁_<img>N products, thereby decreasing the carbon numbers involved in soot formation. This significantly reduces the concentrations of C₅_<img>C₉ molecules and key polycyclic aromatic hydrocarbons (PAHs) precursors like C₂H₂ and C₃H₃. On the other hand, ammonia promotes the ring-opening reactions of six-membered carbon rings at high-temperature conditions, thereby reducing the formation of PAHs precursors. The results show that with the addition of ammonia, six-membered carbon rings tend to convert into seven-membered carbon rings at lower temperatures, while at higher temperatures, they are more likely to transform into three- and five-membered carbon rings. These variations in the transformation of six-membered carbon rings may also affect soot formation. The insights gained from understanding these fundamental chemical reaction mechanisms can guide the development of ethylene-ammonia co-firing systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100454"},"PeriodicalIF":9.6,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142745676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling and optimization of renewable hydrogen systems: A systematic methodological review and machine learning integration","authors":"M.D. Mukelabai,&nbsp;E.R. Barbour ,&nbsp;R.E. Blanchard","doi":"10.1016/j.egyai.2024.100455","DOIUrl":"10.1016/j.egyai.2024.100455","url":null,"abstract":"<div><div>The renewable hydrogen economy is recognized as an integral solution for decarbonizing energy sectors. However, high costs have hindered widespread deployment. One promising way of reducing the costs is optimization. Optimization generally involves finding the configuration of the renewable generation and hydrogen system components that maximizes return on investment. Previous studies have included many aspects into their optimizations, including technical parameters and different costs/socio-economic objective functions, however there is no clear best-practice framework for model development. To address these gaps, this critical review examines the latest development in renewable hydrogen microgrid models and summarizes the best modeling practice. The findings show that advances in machine learning integration are improving solar electricity generation forecasting, hydrogen system simulations, and load profile development, particularly in data-scarce regions. Additionally, it is important to account for electrolyzer and fuel cell dynamics, rather than utilizing fixed performance values. This review also demonstrates that typical meteorological year datasets are better for modeling solar irradiation than first-principle calculations. The practicability of socio-economic objective functions is also assessed, proposing that the more comprehensive Levelized Value Addition (LVA) is best suited for inclusion into models. Best practices for creating load profiles in regions like the Global South are discussed, along with an evaluation of AI-based and traditional optimization methods and software tools. Finally, a new evidence-based multi-criteria decision-making framework integrated with machine learning insights, is proposed to guide decision-makers in selecting optimal solutions based on multiple attributes, offering a more comprehensive and adaptive approach to renewable hydrogen system optimization.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100455"},"PeriodicalIF":9.6,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143131782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing PV feed-in power forecasting through federated learning with differential privacy using LSTM and GRU","authors":"Pascal Riedel ,&nbsp;Kaouther Belkilani ,&nbsp;Manfred Reichert ,&nbsp;Gerd Heilscher ,&nbsp;Reinhold von Schwerin","doi":"10.1016/j.egyai.2024.100452","DOIUrl":"10.1016/j.egyai.2024.100452","url":null,"abstract":"<div><div>Given the inherent fluctuation of photovoltaic (PV) generation, accurately forecasting solar power output and grid feed-in is crucial for optimizing grid operations. Data-driven methods facilitate efficient supply and demand management in smart grids, but predicting solar power remains challenging due to weather dependence and data privacy restrictions. Traditional deep learning (DL) approaches require access to centralized training data, leading to security and privacy risks. To navigate these challenges, this study utilizes federated learning (FL) to forecast feed-in power for the low-voltage grid. We propose a bottom-up, privacy-preserving prediction method using differential privacy (DP) to enhance data privacy for energy analytics on the customer side. This study aims at proving the viability of an enhanced FL approach by employing three years of meter data from three residential PV systems installed in a southern city of Germany, incorporating irradiance weather data for accurate PV power generation predictions. For the experiments, the DL models long short-term memory (LSTM) and gated recurrent unit (GRU) are federated and integrated with DP. Consequently, federated LSTM and GRU models are compared with centralized and local baseline models using rolling 5-fold cross-validation to evaluate their respective performances. By leveraging advanced FL algorithms such as FedYogi and FedAdam, we propose a method that not only predicts sequential energy data with high accuracy, achieving an <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 97.68%, but also adheres to stringent privacy standards, offering a scalable solution for the challenges of smart grids analytics, thus clearly showing that the proposed approach is promising and worth being pursued further.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100452"},"PeriodicalIF":9.6,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142723726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-world validation of safe reinforcement learning, model predictive control and decision tree-based home energy management systems","authors":"Julian Ruddick ,&nbsp;Glenn Ceusters ,&nbsp;Gilles Van Kriekinge ,&nbsp;Evgenii Genov ,&nbsp;Cedric De Cauwer ,&nbsp;Thierry Coosemans ,&nbsp;Maarten Messagie","doi":"10.1016/j.egyai.2024.100448","DOIUrl":"10.1016/j.egyai.2024.100448","url":null,"abstract":"<div><div>Recent advancements in machine learning based energy management approaches, specifically reinforcement learning with a safety layer (<span>OptLayerPolicy</span>) and a metaheuristic algorithm generating a decision tree control policy (<span>TreeC</span>), have shown promise. However, their effectiveness has only been demonstrated in computer simulations. This paper presents the real-world validation of these methods, comparing them against model predictive control and simple rule-based control benchmarks. The experiments were conducted on the electrical installation of four reproductions of residential houses, each with its own battery, photovoltaic, and dynamic load system emulating a non-controllable electrical load and a controllable electric vehicle charger. The results show that the simple rules, <span>TreeC</span>, and model predictive control-based methods achieved similar costs, with a difference of only 0.6%. The reinforcement learning based method, still in its training phase, obtained a cost 25.5% higher to the other methods. Additional simulations show that the costs can be further reduced by using a more representative training dataset for <span>TreeC</span> and addressing errors in the model predictive control implementation caused by its reliance on accurate data from various sources. The <span>OptLayerPolicy</span> safety layer allows safe online training of a reinforcement learning agent in the real world, given an accurate constraint function formulation. The proposed safety layer method remains error-prone; nonetheless, it has been found beneficial for all investigated methods. The <span>TreeC</span> method, which does require building a realistic simulation for training, exhibits the safest operational performance, exceeding the grid limit by only 27.1 Wh compared to 593.9 Wh for reinforcement learning.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"18 ","pages":"Article 100448"},"PeriodicalIF":9.6,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142723725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}