Energy and AI最新文献_第3页

Explainable optimized voting ensemble for photovoltaic power forecasting 光伏发电预测的可解释优化投票集合

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-01-30 DOI: 10.1016/j.egyai.2026.100690

Prince Aduama, Ameena S. Al-Sumaiti, Vikash Kumar Saini, Ruosi Kong

{"title":"Explainable optimized voting ensemble for photovoltaic power forecasting","authors":"Prince Aduama, Ameena S. Al-Sumaiti, Vikash Kumar Saini, Ruosi Kong","doi":"10.1016/j.egyai.2026.100690","DOIUrl":"10.1016/j.egyai.2026.100690","url":null,"abstract":"<div><div>Forecasting photovoltaic (PV) power accurately is essential for enhancing grid reliability and optimizing energy management in renewable power systems. This study proposes an optimized ensemble-based regression framework for improving PV power predictions. Three machine learning models (LSTM, CNN–LSTM, and SVR) are utilized to generate baseline forecasts. To enhance predictive performance, particle swarm optimization (PSO) is utilized for hyperparameter tuning, ensuring optimal model configurations. Furthermore, a weighted ensemble strategy is introduced, where simple voting and grid search-based ensemble voting are compared to refine final predictions. Experimental results demonstrate that the optimized grid search ensemble model achieves superior forecasting accuracy, with MSE of 163.02 kW<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span>, RMSE of 12.77 kW, nRMSE of 1.55%, rRMSE of 6.31%, MAE of 5.26 kW, and <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 0.9976. The scalability and robustness of the model are tested by utilizing data from different regions and a variable cloud analysis respectively, yielding superior results. These findings highlight the critical role of hyperparameter optimization and ensemble weighting in enhancing solar PV power forecasting, offering a robust framework for grid operators and energy planners to improve decision-making in solar-integrated power systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100690"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174325","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}

引用次数: 0

Data-driven landscape scenicness mapping for continental-scale onshore wind resource assessment 基于数据驱动的陆地风力资源评价景观景观制图

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-04-14 DOI: 10.1016/j.egyai.2026.100752

Ruihong Chen , Tristan Pelser , Alena Lohrmann , Jann Michael Weinand , Russell McKenna

{"title":"Data-driven landscape scenicness mapping for continental-scale onshore wind resource assessment","authors":"Ruihong Chen , Tristan Pelser , Alena Lohrmann , Jann Michael Weinand , Russell McKenna","doi":"10.1016/j.egyai.2026.100752","DOIUrl":"10.1016/j.egyai.2026.100752","url":null,"abstract":"<div><div>Visual impacts on scenic landscapes dominate public opposition to onshore wind turbines. Yet wind resource assessments often overlook landscape scenicness due to limited data availability. This study introduces a scalable machine learning framework for generating continental scenicness layers, trained on crowdsourced scenicness ratings from Great Britain and achieving high predictive performance. The resulting scenicness maps are integrated into an onshore wind resource assessment under three landscape preservation scenarios across 29 European countries. We show that prioritizing scenic landscapes in planning can reduce wind generation potential in certain countries by over 60%. However, it only modestly affects the continental median levelized costs of electricity (57 €/MWh and 54 €/MWh under low and high preservation scenarios), while substantially increasing regional costs in scenic mountainous regions such as the Alps and Norway. These findings demonstrate how data-driven approaches can enable socially aware and large-scale energy system planning.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100752"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147797645","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}

引用次数: 0

Battery aging and behavioral pattern identification: A fleet analytics framework for regulatory compliance testing 电池老化和行为模式识别：用于法规遵从性测试的车队分析框架

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2025-12-22 DOI: 10.1016/j.egyai.2025.100669

Robin Saam , Ludwig Hagen Letzig , Jens Grabow , Ralf Benger , Ines Hauer

{"title":"Battery aging and behavioral pattern identification: A fleet analytics framework for regulatory compliance testing","authors":"Robin Saam , Ludwig Hagen Letzig , Jens Grabow , Ralf Benger , Ines Hauer","doi":"10.1016/j.egyai.2025.100669","DOIUrl":"10.1016/j.egyai.2025.100669","url":null,"abstract":"<div><div><em>Battery Electric Vehicle</em>s (BEVs) face stringent regulatory requirements for battery durability, with Euro 7 and <em>Advanced Clean Cars II</em> (ACCII) regulations mandating <em>State of Health</em> (SoH) thresholds of 70% after 8 years or 160,000km and 10 years or 240,000km, respectively. Current testing protocols rely on homogeneous profiles that inadequately represent real-world usage variability, creating recall and over-engineering risks. This paper addresses this gap by proposing a data-driven framework linking usage behavior patterns with battery aging dynamics to identify representative test profiles.</div><div>The framework consists of three integrated steps: (1) predicting battery age and usage at regulatory boundaries through mathematical formulation, (2) estimating SoH using machine learning models including ensemble methods and statistical approaches, and (3) clustering vehicles into groups with homogeneous aging behavior. Unlike existing approaches, it accounts for real-world heterogeneity across diverse usage patterns without assuming predefined load profiles.</div><div>Validation used a large-scale Volkswagen AG fleet of roughly 850,000 vehicles across five countries (China, Germany, Italy, Norway, United States). Analysis across regions demonstrates how usage behavior diversity impacts battery aging, validating the need for behavioral pattern-aware methodologies. The framework was complemented by <em>Center for Advanced Life Cycle Engineering</em> (CALCE) battery cycling validation (192 cells, 24 test groups), achieving Homogeneity 0.9157 and Completeness 0.9509.</div><div>The modular design enables future enhancements including multi-view clustering and advanced SoH prediction models. This framework provides manufacturers a systematic approach to regulatory compliance while reducing unnecessary testing and mitigating over-engineering risk.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100669"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080921","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}

引用次数: 0

Explainable AI for predicting household demand flexibility: Insights from smart meter data and price-based programs 预测家庭需求灵活性的可解释人工智能：来自智能电表数据和基于价格的程序的见解

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-01-21 DOI: 10.1016/j.egyai.2026.100686

Santiago Bañales , Raquel Dormido , Natividad Duro

{"title":"Explainable AI for predicting household demand flexibility: Insights from smart meter data and price-based programs","authors":"Santiago Bañales , Raquel Dormido , Natividad Duro","doi":"10.1016/j.egyai.2026.100686","DOIUrl":"10.1016/j.egyai.2026.100686","url":null,"abstract":"<div><div>Unlocking Demand‐Side Flexibility (DSF) at scale is essential for integrating variable renewables and electrified end-uses. We develop a scalable, explainable-AI framework to assess the predictability and drivers of household responsiveness to price-based programs using only data typically available to utilities (smart meters, basic weather, limited socio-economic tags). Using the public Low Carbon London Time-of-Use (ToU) pilot, we first estimate responsiveness with Least Absolute Shrinkage and Selection Operator (LASSO) at both aggregated and household levels—overall and by hour—to quantify effect sizes and heterogeneity. We then train Gradient-Boosting (GB) models and apply SHapley Additive exPlanations (SHAP) to assess the hierarchy and direction of drivers of flexibility. Results show statistically significant but moderate average responses with wide dispersion across households and time-of-day, including a significant percentage of counter-intuitive reactions to price. Features capturing unexplained variability in hourly and daily load (e.g., dispersion measures of residual components) are the strongest positive predictors of flexibility, whereas seasonality/predictability indicators (autocorrelation and seasonal strength) are neutral or negative. SHAP dependence plots reveal clear thresholds, breakpoints, and saturation effects, underscoring the nonlinearity of behavioral response. Because the feature set is derived from routinely collected data, the approach is replicable and operationally practical. The findings enable data-driven targeting of high-potential households and support the design of digital orchestration platforms for near-time demand response, informing tariff design, aggregator strategies, and regulatory guidance for market-based DSF.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100686"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081013","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}

引用次数: 0

From peak shaving to stress redistribution: Emergent heatwave resilience through weather-informed decentralized reinforcement learning 从调峰到应力再分配：通过天气信息分散强化学习的紧急热浪恢复力

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-04-24 DOI: 10.1016/j.egyai.2026.100762

Vahid M. Nik

{"title":"From peak shaving to stress redistribution: Emergent heatwave resilience through weather-informed decentralized reinforcement learning","authors":"Vahid M. Nik","doi":"10.1016/j.egyai.2026.100762","DOIUrl":"10.1016/j.egyai.2026.100762","url":null,"abstract":"<div><div>Heatwaves increasingly challenge urban electricity networks by intensifying and reshaping cooling demand over extended periods. This study investigates how weather-informed Adaptive Reinforcement Learning for Energy Management (ARLEM), a decentralized model-free control framework, redistributes cooling demand and manages systems stress during sustained heatwaves, focusing on the roles of forecast horizon, policy-update frequency and action-selection constraints in determining climate-resilient control performance. ARLEM is evaluated for a representative residential neighbourhood in Madrid consisting of four building archetypes and simulated under near-future summer climate conditions derived from an ensemble of regional climate projections. A broad range of control configurations is explored, including predictive and non-predictive learning, different forecast horizons, policy-update schedules, and action-selection constraints. System behaviour is assessed using stress-oriented indicators that capture demand flexibility, adaptive responsiveness, and amplification of demand under extreme conditions.</div><div>Results show that predictive ARLEM operates as a distribution-shaping controller rather than a peak-only optimizer. During heatwaves, predictive ARLEM reduces neighbourhood-scale mean cooling demand by approximately 10–15% while lowering high-quantile demand levels by 80–130 kWh, without destabilizing indoor thermal conditions. Flexibility is allocated unevenly across buildings, allowing heterogeneous responses that collectively reduce system stress without synchronized rebound effects. The findings demonstrate that resilience in decentralized adaptive control emerges from coherent alignment between anticipation, adaptation speed, and action constraints. By managing sustained stress rather than merely suppressing short-lived peaks, weather-informed ARLEM provides a scalable, privacy-preserving, and context-aware framework for neighbourhood-scale energy management under increasingly severe climate conditions.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100762"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147797587","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}

引用次数: 0

Machine learning tabulation of thermochemistry: Transfer learning for non-adiabatic flames 热化学的机器学习制表：非绝热火焰的迁移学习

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-04-20 DOI: 10.1016/j.egyai.2026.100746

Alireza Darzi, W.P. Jones, Stelios Rigopoulos

{"title":"Machine learning tabulation of thermochemistry: Transfer learning for non-adiabatic flames","authors":"Alireza Darzi, W.P. Jones, Stelios Rigopoulos","doi":"10.1016/j.egyai.2026.100746","DOIUrl":"10.1016/j.egyai.2026.100746","url":null,"abstract":"<div><div>Artificial Neural Networks (ANNs) are a powerful tool to accelerate the calculation of the thermochemistry source term in turbulent reacting flow simulations. However, their capacity for transfer learning, i.e. their ability to predict flames different from the problems they were trained on, depends on the way training data are generated. In the Hybrid Flamelet/Random Data (HFRD) approach (Ding et al., Combust. Flame 231:111493, 2021), data are generated with a generic approach that combines canonical laminar flames with randomisation, and the resulting dataset enables training ANNs that can generalise to wide families of turbulent flames. However, in many applications such as gas turbines, non-adiabatic processes such as wall heat loss result in an extended composition space. In the present work, we introduce a temperature-profile-control strategy for data generation that enlarges the HFRD composition space coverage by employing a set of one-dimensional burner-stabilised flames with specified temperature profile at various flow rates and mixture fractions. The resulting ANNs are applied first to a one-dimensional freely-propagating CH<sub>4</sub>/H<sub>2</sub>-air premixed flame, to demonstrate transfer to a laminar flame problem with a different fuel inlet. Subsequently, the overall approach is applied to the simulation of the model gas turbine combustor PRECCINSTA with a Large Eddy Simulation-Probability Density Function (LES-PDF) method. It must be noted that the machine learning approach is applicable to any turbulence-chemistry interaction model involving real-time integration of kinetics. Results from the ANN-accelerated simulations show very good agreement with simulations involving direct integration of the kinetics.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100746"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147849842","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}

引用次数: 0

Trustworthy and explainable deep reinforcement learning for safe and energy-efficient process control: A use case in industrial compressed air systems 用于安全和节能过程控制的可信赖且可解释的深度强化学习：工业压缩空气系统中的一个用例

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-02-05 DOI: 10.1016/j.egyai.2026.100685

Vincent Bezold , Patrick Wagner , Jakob Hofmann , Marco Huber , Alexander Sauer

{"title":"Trustworthy and explainable deep reinforcement learning for safe and energy-efficient process control: A use case in industrial compressed air systems","authors":"Vincent Bezold , Patrick Wagner , Jakob Hofmann , Marco Huber , Alexander Sauer","doi":"10.1016/j.egyai.2026.100685","DOIUrl":"10.1016/j.egyai.2026.100685","url":null,"abstract":"<div><div>This paper presents a trustworthy reinforcement learning approach for the control of industrial compressed air systems. We develop a framework that enables safe and energy-efficient operation under realistic boundary conditions and introduce a multi-level explainability pipeline combining input perturbation tests, gradient-based sensitivity analysis, and SHAP (SHapley Additive exPlanations) feature attribution. An empirical evaluation across multiple compressor configurations shows that the learned policy is physically plausible, anticipates future demand, and consistently respects system boundaries. Compared to the installed industrial controller, the proposed approach reduces unnecessary overpressure and achieves energy savings of approximately 4% without relying on explicit physics models. The results further indicate that system pressure and forecast information dominate policy decisions, while compressor-level inputs play a secondary role. Overall, the combination of efficiency gains, predictive behavior, and transparent validation supports the trustworthy deployment of reinforcement learning in industrial energy systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100685"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174326","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}

引用次数: 0

Power system operational reliability evaluation with retrieval-augmented generation enhanced large language model 基于检索增强生成的大语言模型的电力系统运行可靠性评估

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-02-02 DOI: 10.1016/j.egyai.2026.100688

Yuheng Cheng , Huan Zhao , Dejun Xiang , Zhengwen Zhang , Guolong Liu , Yanli Liu , Junhua Zhao , Xinlei Cai

{"title":"Power system operational reliability evaluation with retrieval-augmented generation enhanced large language model","authors":"Yuheng Cheng , Huan Zhao , Dejun Xiang , Zhengwen Zhang , Guolong Liu , Yanli Liu , Junhua Zhao , Xinlei Cai","doi":"10.1016/j.egyai.2026.100688","DOIUrl":"10.1016/j.egyai.2026.100688","url":null,"abstract":"<div><div>Operational reliability evaluation of power grid dispatch is crucial for ensuring power system security and stability. Traditional evaluation involves system state assessment and operation regulations verification, with the latter relying heavily on dispatcher expertise, leading to inefficiencies and accuracy limitations. While rule-based automation has partially addressed these challenges, generalization to frequently updated regulations remains problematic. This paper proposes a Retrieval-Augmented Generation enhanced Large Language Model framework for automated operation regulations evaluation in power grid dispatch. The framework features natural language understanding capabilities for accurate semantic interpretation of dispatch regulations and supports real-time knowledge updates without additional model training. A Hierarchical Document Retrieval method improves retrieval precision by leveraging document structure, while an Operations AutoPrompt Generation technique automatically converts dispatch operations into optimized queries. Experimental validation using simulation data and real operational data from the Guangdong Power Grid demonstrates that the proposed method achieves an average evaluation accuracy of 90% across multiple international regulation datasets, with compliance accuracy reaching 93% on the Italian Grid Code. The framework processes evaluation queries in 5.6 to 7.0 s, meeting practical requirements for dispatch decision support. These results demonstrate the framework’s strong generalization ability and practical applicability for real-world power grid operational reliability evaluation.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100688"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174324","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}

引用次数: 0

Accelerated computing of dynamic carbon emission factor for large-scale power grid using conjugate gradient squared method with p-norm preconditioner 基于p范数预调节器的共轭梯度平方法加速大规模电网动态碳排放因子计算

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2026-03-27 DOI: 10.1016/j.egyai.2026.100730

Bin Wang, Julong Chen, Chen Luo, Xuepeng Mou, Shiping Yang

{"title":"Accelerated computing of dynamic carbon emission factor for large-scale power grid using conjugate gradient squared method with p-norm preconditioner","authors":"Bin Wang, Julong Chen, Chen Luo, Xuepeng Mou, Shiping Yang","doi":"10.1016/j.egyai.2026.100730","DOIUrl":"10.1016/j.egyai.2026.100730","url":null,"abstract":"<div><div>Amid the global drive for energy decarbonization, power systems—accounting for over 40% of energy-sector carbon emissions—demand precise dynamic carbon tracking and low-carbon scheduling tools. Conventional methods for calculating electricity carbon emission factors (CEFs) are ill-suited for large-scale grids due to critical limitations: direct inversion suffers from O(n³) complexity and poor numerical stability, while iterative approaches such as the least squares QR method converge slowly, and the conjugate gradient squared method may diverge or oscillate under severe ill-conditioning, failing to meet real-time requirements. To address these issues, this study presents a p-norm preconditioner-boosted conjugate gradient least squares (P-CGS) approach. A multi-norm adaptive preconditioning strategy (1 ≤ p ≤ ∞) is developed to improve matrix conditioning, integrated with a dynamic initial-value transfer mechanism that leverages temporal load correlations. This enables efficient computation of multi-period dynamic CEF surfaces with relative residuals controlled below 10⁻⁸, aligning with precision standards for high-fidelity power flow and electricity-carbon coupling studies. Meanwhile, the comprehensive validations are conducted on 200-, 500-, and 2000-bus grids, and performances of p-norm preconditioner-boosted conjugate gradient least squares are revealed compared with bi-conjugate gradient, generalized minimal residual, Least Squares QR decomposition algorithm (LSQR), CGS, and direct inversion methods. The results illustrate that preconditioner-boosted conjugate gradient least squares can achieve smaller computation time and iterations under the designated relative residuals compared to others mentioned above. This work supports accurate carbon responsibility allocation, carbon market compliance, and international carbon tariff negotiations, providing robust technical foundations for power system decarbonization.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100730"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147797647","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}

引用次数: 0

Combining values and images in deep learning models for time series forecasting: An electricity market case study 结合时间序列预测的深度学习模型中的值和图像：电力市场案例研究

IF 9.6

Energy and AI Pub Date : 2026-05-01 Epub Date: 2025-12-22 DOI: 10.1016/j.egyai.2025.100673

S. Sáez-Bombín , L. Melgar-García , A. Troncoso

{"title":"Combining values and images in deep learning models for time series forecasting: An electricity market case study","authors":"S. Sáez-Bombín , L. Melgar-García , A. Troncoso","doi":"10.1016/j.egyai.2025.100673","DOIUrl":"10.1016/j.egyai.2025.100673","url":null,"abstract":"<div><div>Most machine learning algorithms for time series forecasting focus on the real values of the time series, ignoring the information that can be found in its graphical representation. On the other hand, the results obtained by deep learning models in terms of extracting the relationships and patterns hidden in the data motivate the development of hybrid or multimodal models in which both the temporal and graphical information of the time series are used. This work explores the combination of this information in the field of deep learning applied to time series forecasting. Thus, this paper proposes a hybrid deep learning model based on the combination of time series images and their real values for time series forecasting. First, a deep convolutional neural network architecture obtains an initial approximation to the time series predictions from images. Secondly, these predictions along with the actual values of the time series feed a recurrent neural network based on gate recurrent units including attention mechanisms to obtain the final forecasts. Results using three electricity-related datasets have been reported, showing that lower errors are obtained with a shorter training time when considering the graphical representation of the time series together with attention mechanisms in the recurrent networks.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"24 ","pages":"Article 100673"},"PeriodicalIF":9.6,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080922","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}

引用次数: 0