Applied Energy最新文献_第5页

Projecting solar peak hours in southern Spain using temperature-based machine learning models until 2100 使用基于温度的机器学习模型预测西班牙南部到2100年的太阳能高峰时间

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-17 DOI: 10.1016/j.apenergy.2026.127395

Juan Antonio Bellido-Jiménez , Javier Estévez , Amanda P. García-Marín

The present work firstly assessed the performance of several temperature-based machine learning models for estimating daily solar radiation (Rs) in Andalusia (Southern Spain) using meteorological data from 122 weather stations. The most accurate models were used to then obtain solar peak hours (SPH) projections up to 2100, providing a standardized measure of the solar energy available per year. All the models outperformed the empirical Hargreaves-Samani method at all locations, obtaining the best results, in general, using multilayer perceptron model.

To estimate Rs and project SPH values after validating the models, daily future gridded projections of temperature data were used for different climate change scenarios. Under moderate scenario, average annual SPH values increased from ∼1850 kWh/m² year in the period 2024–2030 to ∼1950 kWh/m² year by 2100. The high-emission scenario exhibited even more pronounced growth, with SPH exceeding 2000 kWh/m² year. The tendency analysis confirmed, in general, a significant positive trend in SPH values across most of Andalusia. However, some coastal areas showed minimal or even negative SPH trends.

This study highlights the increasing solar energy potential in Southern Spain over the coming decades, supporting the transition to renewable energy. The models demonstrate the feasibility of temperature-based approaches for estimating these variables (Rs, SPH), which can help to identify optimal locations for solar energy infrastructure. The high-resolution SPH projections evaluated, and the methodology used, are valuable tools for policymakers and energy planners, allowing informed decision-making in the current context of climate variability and the growing energy demand.

目前的工作首先评估了几种基于温度的机器学习模型的性能，这些模型使用来自122个气象站的气象数据来估计安达卢西亚（西班牙南部）的日太阳辐射（Rs）。然后使用最精确的模型来获得到2100年的太阳峰值时间（SPH）预测，提供每年可用太阳能的标准化测量。所有模型在所有位置都优于经验Hargreaves-Samani方法，总体而言，使用多层感知器模型获得了最好的结果。为了在模型验证后估算Rs和预估SPH值，使用了不同气候变化情景下的逐日温度数据网格预估。在中等情景下，平均年SPH值从2024-2030年的~ 1850 kWh/m2年增加到2100年的~ 1950 kWh/m2年。高排放情景的增长更为明显，SPH年超过2000 kWh/m2。趋势分析证实，总体而言，安达卢西亚大部分地区的SPH值呈显著的正趋势。然而，部分沿海地区的保护海港指数呈轻微甚至负的趋势。这项研究强调了未来几十年西班牙南部太阳能潜力的增加，支持向可再生能源的过渡。这些模型证明了基于温度的方法估算这些变量（Rs， SPH）的可行性，这有助于确定太阳能基础设施的最佳位置。所评估的高分辨率SPH预测和使用的方法对政策制定者和能源规划者来说是有价值的工具，可以在当前气候变化和不断增长的能源需求的背景下做出明智的决策。

{"title":"Projecting solar peak hours in southern Spain using temperature-based machine learning models until 2100","authors":"Juan Antonio Bellido-Jiménez , Javier Estévez , Amanda P. García-Marín","doi":"10.1016/j.apenergy.2026.127395","DOIUrl":"10.1016/j.apenergy.2026.127395","url":null,"abstract":"<div><div>The present work firstly assessed the performance of several temperature-based machine learning models for estimating daily solar radiation (Rs) in Andalusia (Southern Spain) using meteorological data from 122 weather stations. The most accurate models were used to then obtain solar peak hours (SPH) projections up to 2100, providing a standardized measure of the solar energy available per year. All the models outperformed the empirical Hargreaves-Samani method at all locations, obtaining the best results, in general, using multilayer perceptron model.</div><div>To estimate Rs and project SPH values after validating the models, daily future gridded projections of temperature data were used for different climate change scenarios. Under moderate scenario, average annual SPH values increased from ∼1850 kWh/m<sup>2</sup> year in the period 2024–2030 to ∼1950 kWh/m<sup>2</sup> year by 2100. The high-emission scenario exhibited even more pronounced growth, with SPH exceeding 2000 kWh/m<sup>2</sup> year. The tendency analysis confirmed, in general, a significant positive trend in SPH values across most of Andalusia. However, some coastal areas showed minimal or even negative SPH trends.</div><div>This study highlights the increasing solar energy potential in Southern Spain over the coming decades, supporting the transition to renewable energy. The models demonstrate the feasibility of temperature-based approaches for estimating these variables (Rs, SPH), which can help to identify optimal locations for solar energy infrastructure. The high-resolution SPH projections evaluated, and the methodology used, are valuable tools for policymakers and energy planners, allowing informed decision-making in the current context of climate variability and the growing energy demand.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127395"},"PeriodicalIF":11.0,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146024770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A risk-averse multi-timescale stochastic integration framework for a community multi-energy system under correlated uncertainties 关联不确定性下社区多能系统的风险规避多时间尺度随机积分框架

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-17 DOI: 10.1016/j.apenergy.2026.127364

Chaoyu Jia , Huiyuan Wang , Yan Cao , Hongjie Jia , Yunfei Mu , Xiaodan Yu

Diverse correlated uncertainties may bring operational risks and lead to real-time energy shortages of the community multi-energy system (CMES), necessitating efficient coordination of heterogeneous resources through integrated optimization models. To address this problem, a risk-averse multi-timescale stochastic system integration framework for CMES under various correlated uncertainties is proposed in this paper. Operation optimization for the CMES is dynamically coordinated across different timescales, including day-ahead phase optimization (OPT2-DA), intraday phase optimization (OPT2-ID) and real-time local optimization (OPT1). In OPT2-DA, a conditional value at risk (CVaR)-based three-stage stochastic programming (SP) model is established to handle multi-dimensional correlated uncertainties and mitigate potential economic risks. The day-ahead strategy, including unit start-up status, day-ahead power purchase and energy storage system (ESS) dispatch, is optimized. In the subsequent OPT2-ID, as part of uncertainties gradually unfold, a two-stage SP model with rolling optimization is established, addressing prediction errors within the operation horizon. The intraday strategy, including unit output and intraday power purchase, is continually optimized for updating. Building on day-ahead and intraday strategies, OPT1 uses real-time rolling optimization to dynamically adjust operation strategies, ensuring real-time balance of energy supply and demand to prevent shortages. To model diverse correlated uncertainties from renewable energy sources (RESs), multi-energy loads and energy prices across different timescales, a new scenario generation and reduction method that combines Cholesky decomposition-based Latin hypercubic sampling (CD-LHS) and Gaussian mixture model (GMM)-based clustering is developed, forming a three-layer scenario tree. Finally, case studies verify the effectiveness of the proposed method, presenting a lower CVaR than the traditional risk-neutral method and achieving 100% energy supply adequacy.

多种相互关联的不确定性可能带来运行风险，导致社区多能源系统（CMES）的实时能源短缺，需要通过集成优化模型对异构资源进行高效协调。针对这一问题，本文提出了一种风险规避型多时间尺度CMES随机系统集成框架。cme的运行优化在不同时间尺度上是动态协调的，包括日前阶段优化（OPT2-DA）、日内阶段优化（OPT2-ID）和实时局部优化（OPT1）。在OPT2-DA中，建立了一种基于条件风险值（CVaR）的三阶段随机规划（SP）模型，以处理多维相关不确定性，降低潜在经济风险。对机组启动状态、日前购电和储能系统调度等日前策略进行了优化。在后续的OPT2-ID中，随着部分不确定因素的逐渐展开，建立了滚动优化的两阶段SP模型，解决了作业范围内的预测误差。日内策略，包括单位输出和日内购电，不断优化更新。OPT1以日前策略和日内策略为基础，通过实时滚动优化，动态调整运营策略，确保能源供需的实时平衡，防止能源短缺。为了模拟不同时间尺度下可再生能源（RESs）、多能负荷和能源价格的多种相关不确定性，提出了一种基于Cholesky分解的拉丁超立方采样（CD-LHS）和高斯混合模型（GMM）聚类相结合的情景生成与约简方法，形成了三层情景树。最后，通过案例研究验证了该方法的有效性，CVaR低于传统的风险中性方法，实现了100%的能源供应充足性。

{"title":"A risk-averse multi-timescale stochastic integration framework for a community multi-energy system under correlated uncertainties","authors":"Chaoyu Jia , Huiyuan Wang , Yan Cao , Hongjie Jia , Yunfei Mu , Xiaodan Yu","doi":"10.1016/j.apenergy.2026.127364","DOIUrl":"10.1016/j.apenergy.2026.127364","url":null,"abstract":"<div><div>Diverse correlated uncertainties may bring operational risks and lead to real-time energy shortages of the community multi-energy system (CMES), necessitating efficient coordination of heterogeneous resources through integrated optimization models. To address this problem, a risk-averse multi-timescale stochastic system integration framework for CMES under various correlated uncertainties is proposed in this paper. Operation optimization for the CMES is dynamically coordinated across different timescales, including day-ahead phase optimization (OPT2-DA), intraday phase optimization (OPT2-ID) and real-time local optimization (OPT1). In OPT2-DA, a conditional value at risk (CVaR)-based three-stage stochastic programming (SP) model is established to handle multi-dimensional correlated uncertainties and mitigate potential economic risks. The day-ahead strategy, including unit start-up status, day-ahead power purchase and energy storage system (ESS) dispatch, is optimized. In the subsequent OPT2-ID, as part of uncertainties gradually unfold, a two-stage SP model with rolling optimization is established, addressing prediction errors within the operation horizon. The intraday strategy, including unit output and intraday power purchase, is continually optimized for updating. Building on day-ahead and intraday strategies, OPT1 uses real-time rolling optimization to dynamically adjust operation strategies, ensuring real-time balance of energy supply and demand to prevent shortages. To model diverse correlated uncertainties from renewable energy sources (RESs), multi-energy loads and energy prices across different timescales, a new scenario generation and reduction method that combines Cholesky decomposition-based Latin hypercubic sampling (CD-LHS) and Gaussian mixture model (GMM)-based clustering is developed, forming a three-layer scenario tree. Finally, case studies verify the effectiveness of the proposed method, presenting a lower CVaR than the traditional risk-neutral method and achieving 100% energy supply adequacy.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127364"},"PeriodicalIF":11.0,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A novel day-ahead bidding method considering real-time operational characteristic and strategy for wind-hydro hybrid power systems 一种考虑风力-水力混合发电系统实时运行特性和策略的日前竞价方法

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-17 DOI: 10.1016/j.apenergy.2026.127350

Chunyang Lai, Behzad Kazemtabrizi

Two-stage stochastic model is the state-of-the-art method for making the day-ahead bid for wind–hydro hybrid power systems (WHHPS) considering both the day-ahead and the real-time balancing markets. However, the real-time operational characteristic and strategy of WHHPS are ignored in current studies which limits their applications in day-ahead bidding. To fill this gap, in this paper, a novel day-ahead bidding method considering real-time operational characteristic and strategy for WHHPS is proposed. First, using Kantorovich distance, a scenario tree generation algorithm is introduced to capture the stage-wise unfolding nature of the wind power and balancing market price realisations in the real-time operation. Then, a bilevel bidding method is proposed, whose upper level aims to maximise the expected income while managing the income risk of the WHHPS, and the lower level operates the WHHPS considering the real-time operational characteristic and strategy. Due to the chronological order in real-time operation, the lower level contains sequentially nested problems which makes the whole model difficult to solve. For this reason, a weighted reformulation method is proposed to equally reformulate the sequentially nested problems in the lower level into a single LP problem. Finally, the bilevel model is reformulated as MILP. Comparisons are made with the current state-of-the-art models in a case study. The results show that the bidding strategies derived from the conventional two-stage stochastic bidding model tend to be overly optimistic regarding both the expected income and the associated risk. Moreover, using CVaR may fail to accurately reflect the operator’s risk preference in bidding. In comparison, the proposed method establishes a monotonic decreasing relationship between income risk and the coefficient

P

, providing a direct approach for risk management.

两阶段随机模型是目前最先进的既考虑了日前市场又考虑了实时平衡市场的风电混合发电系统日前报价方法。然而，目前的研究忽略了WHHPS的实时运行特性和策略，限制了其在日前竞价中的应用。为了填补这一空白，本文提出了一种考虑WHHPS实时运行特性和策略的日前竞价方法。首先，利用Kantorovich距离，引入一种场景树生成算法来捕捉风电的分阶段展开特性，并在实时运行中平衡市场价格实现。在此基础上，提出了一种双层投标方法，其上层以期望收益最大化为目标，同时对WHHPS的收益风险进行管理，下层则考虑WHHPS的实时运行特点和策略来运营WHHPS。由于实时操作是按时间顺序进行的，底层包含了顺序嵌套的问题，使得整个模型难以求解。为此，提出了一种加权重表述方法，将较低层次的顺序嵌套问题等价地重新表述为单个LP问题。最后，将双层模型重新表述为MILP。在一个案例研究中，与当前最先进的模型进行了比较。研究结果表明，传统的两阶段随机竞价模型推导出的竞价策略在预期收益和相关风险方面都过于乐观。此外，使用CVaR可能无法准确反映运营商在投标中的风险偏好。相比之下，该方法建立了收入风险与系数P之间的单调递减关系，为风险管理提供了一种直接的方法。

{"title":"A novel day-ahead bidding method considering real-time operational characteristic and strategy for wind-hydro hybrid power systems","authors":"Chunyang Lai, Behzad Kazemtabrizi","doi":"10.1016/j.apenergy.2026.127350","DOIUrl":"10.1016/j.apenergy.2026.127350","url":null,"abstract":"<div><div>Two-stage stochastic model is the state-of-the-art method for making the day-ahead bid for wind–hydro hybrid power systems (WHHPS) considering both the day-ahead and the real-time balancing markets. However, the real-time operational characteristic and strategy of WHHPS are ignored in current studies which limits their applications in day-ahead bidding. To fill this gap, in this paper, a novel day-ahead bidding method considering real-time operational characteristic and strategy for WHHPS is proposed. First, using Kantorovich distance, a scenario tree generation algorithm is introduced to capture the stage-wise unfolding nature of the wind power and balancing market price realisations in the real-time operation. Then, a bilevel bidding method is proposed, whose upper level aims to maximise the expected income while managing the income risk of the WHHPS, and the lower level operates the WHHPS considering the real-time operational characteristic and strategy. Due to the chronological order in real-time operation, the lower level contains sequentially nested problems which makes the whole model difficult to solve. For this reason, a weighted reformulation method is proposed to equally reformulate the sequentially nested problems in the lower level into a single LP problem. Finally, the bilevel model is reformulated as MILP. Comparisons are made with the current state-of-the-art models in a case study. The results show that the bidding strategies derived from the conventional two-stage stochastic bidding model tend to be overly optimistic regarding both the expected income and the associated risk. Moreover, using CVaR may fail to accurately reflect the operator’s risk preference in bidding. In comparison, the proposed method establishes a monotonic decreasing relationship between income risk and the coefficient <span><math><mrow><mi>P</mi></mrow></math></span>, providing a direct approach for risk management.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127350"},"PeriodicalIF":11.0,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146024771","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adaptive forecasting of photovoltaic power based on dual-type models’ ensemble and online error correction 基于双型模型集成和在线误差校正的光伏发电自适应预测

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-17 DOI: 10.1016/j.apenergy.2026.127397

Haipeng Liu, Hong Wu, Huaiping Jin, Yanping He

Solar energy has become an important part of clean energy. Reliable photovoltaic (PV) power forecasting can effectively increase solar energy utilization and reduce costs, which is crucial to the optimal scheduling of PV power generation. However, accurate forecasting of PV power is facing enormous challenges owing to the inherent intermittency and volatility of solar energy. To address the aforementioned issues, a novel adaptive forecasting method for PV power based on dual-type models’ ensemble and online error correction (DMEOEC) is proposed. Firstly, to improve the quality of the basic model, DMEOEC introduces the temporal-aware block to construct a regularity ensemble module. Besides, it also employs the fuzzy entropy (FE) method to refine the generation of small sample data, thereby building a specificity ensemble module. Secondly, an adaptive Bayesian aggregation algorithm achieves the selective fusion of basic models in different modules. Thirdly, DMEOEC establishes two hierarchical weights for the ensemble modules to enhance the robustness and adaptability of the model in a dynamic environment, and the errors obtained from predictions are used to train an online incremental learning model to correct the hierarchical weights in real-time. Lastly, DMEOEC adopts a temporal prioritized experience replay (TPER) and buffer mechanism to cope with model performance degradation due to concept drift and catastrophic forgetting. Compared with traditional methods, DMEOEC not only captures real-time PV power information but also adapts to environmental changes, thus significantly enhancing the accuracy and reliability of PV power forecasting. The effectiveness and superiority of the proposed DMEOEC method are verified using five real PV power plant datasets.

太阳能已成为清洁能源的重要组成部分。可靠的光伏发电功率预测可以有效提高太阳能利用率，降低成本，对光伏发电的优化调度至关重要。然而，由于太阳能固有的间歇性和波动性，光伏发电的准确预测面临着巨大的挑战。针对上述问题，提出了一种基于双模型集成和在线误差校正（DMEOEC）的光伏发电功率自适应预测方法。首先，为了提高基本模型的质量，DMEOEC引入了时间感知块来构建规则集成模块；此外，还采用模糊熵（FE）方法对小样本数据的生成进行细化，从而构建特异性集成模块。其次，采用自适应贝叶斯聚合算法实现了不同模块中基本模型的选择性融合；第三，DMEOEC为集成模块建立了两个层次权值，以增强模型在动态环境中的鲁棒性和适应性，并利用预测得到的误差训练在线增量学习模型，实时修正层次权值。最后，DMEOEC采用时间优先体验重放（TPER）和缓冲机制来应对概念漂移和灾难性遗忘导致的模型性能下降。与传统方法相比，DMEOEC既能实时捕获光伏发电信息，又能适应环境变化，显著提高了光伏发电预测的准确性和可靠性。通过5个实际的光伏电站数据集验证了所提出的DMEOEC方法的有效性和优越性。

{"title":"Adaptive forecasting of photovoltaic power based on dual-type models’ ensemble and online error correction","authors":"Haipeng Liu, Hong Wu, Huaiping Jin, Yanping He","doi":"10.1016/j.apenergy.2026.127397","DOIUrl":"10.1016/j.apenergy.2026.127397","url":null,"abstract":"<div><div>Solar energy has become an important part of clean energy. Reliable photovoltaic (PV) power forecasting can effectively increase solar energy utilization and reduce costs, which is crucial to the optimal scheduling of PV power generation. However, accurate forecasting of PV power is facing enormous challenges owing to the inherent intermittency and volatility of solar energy. To address the aforementioned issues, a novel adaptive forecasting method for PV power based on dual-type models’ ensemble and online error correction (DMEOEC) is proposed. Firstly, to improve the quality of the basic model, DMEOEC introduces the temporal-aware block to construct a regularity ensemble module. Besides, it also employs the fuzzy entropy (FE) method to refine the generation of small sample data, thereby building a specificity ensemble module. Secondly, an adaptive Bayesian aggregation algorithm achieves the selective fusion of basic models in different modules. Thirdly, DMEOEC establishes two hierarchical weights for the ensemble modules to enhance the robustness and adaptability of the model in a dynamic environment, and the errors obtained from predictions are used to train an online incremental learning model to correct the hierarchical weights in real-time. Lastly, DMEOEC adopts a temporal prioritized experience replay (TPER) and buffer mechanism to cope with model performance degradation due to concept drift and catastrophic forgetting. Compared with traditional methods, DMEOEC not only captures real-time PV power information but also adapts to environmental changes, thus significantly enhancing the accuracy and reliability of PV power forecasting. The effectiveness and superiority of the proposed DMEOEC method are verified using five real PV power plant datasets.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127397"},"PeriodicalIF":11.0,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975801","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Internal temperature sensing for lithium-ion battery under ultrahigh-rate discharge conditions using physics-informed neural network 基于物理信息神经网络的超高倍率放电条件下锂离子电池内部温度传感

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2026.127394

Yupeng Si , Xing Zhou , Yu Wang , Tao Zhang , Yukang Xiao , Yajie Liu

In modern military applications such as directional energy and electromagnetic emission, lithium-ion batteries (LIBs) need to be operated under ultrahigh-rate discharge conditions, where the temperature difference between the core and the surface of cell exceed tens of degrees celsius. Such thermal gradients pose great difficulties for the conventional thermal management techniques that rely on cell surface temperature, leading to severe underestimation of the maximum temperature and non-negligible control delay in thermal regulation, thereby increasing the risk of overheating. To address this critical challenge, this study proposes a self-adaptive multi-magnitude physics-informed neural network (SA-MMPINN) framework for internal temperature sensing. This framework integrates thermal model into neural networks architectures, enabling high-accuracy sense of internal temperature based solely on surface temperature while simultaneously identifying convective heat transfer coefficients to adapt to different heat dissipation conditions. Experimental validation using instrumented cells with embedded thermocouples shows that the framework can achieve root mean square error below 0.76 °C for core temperature sensing under 20C discharge conditions with temperature ranging from −5 °C to 35 °C. Meanwhile, the convective heat transfer coefficient can be identified with over 97% accuracy. The proposed SA-MMPINN framework enables accurate sensing for the internal temperature evolution of LIBs, which is crucial for informing the design of next-generation intelligent thermal management systems.

在定向能和电磁发射等现代军事应用中，锂离子电池（LIBs）需要在超高倍率放电条件下工作，电池芯和电池表面之间的温差超过数十摄氏度。这种热梯度给依赖于电池表面温度的传统热管理技术带来了很大的困难，导致对最高温度的严重低估和热调节中不可忽略的控制延迟，从而增加了过热的风险。为了解决这一关键挑战，本研究提出了一种用于内部温度传感的自适应多量级物理信息神经网络（SA-MMPINN）框架。该框架将热模型集成到神经网络架构中，实现仅基于表面温度的高精度内部温度感知，同时识别对流换热系数，以适应不同的散热条件。使用内置热电偶的仪器电池进行实验验证表明，该框架在20C放电条件下，温度范围为- 5°C至35°C，核心温度传感的均方根误差低于0.76°C。同时，对流换热系数的识别准确率达到97%以上。所提出的SA-MMPINN框架能够准确地感知lib的内部温度演变，这对于下一代智能热管理系统的设计至关重要。

{"title":"Internal temperature sensing for lithium-ion battery under ultrahigh-rate discharge conditions using physics-informed neural network","authors":"Yupeng Si , Xing Zhou , Yu Wang , Tao Zhang , Yukang Xiao , Yajie Liu","doi":"10.1016/j.apenergy.2026.127394","DOIUrl":"10.1016/j.apenergy.2026.127394","url":null,"abstract":"<div><div>In modern military applications such as directional energy and electromagnetic emission, lithium-ion batteries (LIBs) need to be operated under ultrahigh-rate discharge conditions, where the temperature difference between the core and the surface of cell exceed tens of degrees celsius. Such thermal gradients pose great difficulties for the conventional thermal management techniques that rely on cell surface temperature, leading to severe underestimation of the maximum temperature and non-negligible control delay in thermal regulation, thereby increasing the risk of overheating. To address this critical challenge, this study proposes a self-adaptive multi-magnitude physics-informed neural network (SA-MMPINN) framework for internal temperature sensing. This framework integrates thermal model into neural networks architectures, enabling high-accuracy sense of internal temperature based solely on surface temperature while simultaneously identifying convective heat transfer coefficients to adapt to different heat dissipation conditions. Experimental validation using instrumented cells with embedded thermocouples shows that the framework can achieve root mean square error below 0.76 °C for core temperature sensing under 20C discharge conditions with temperature ranging from −5 °C to 35 °C. Meanwhile, the convective heat transfer coefficient can be identified with over 97% accuracy. The proposed SA-MMPINN framework enables accurate sensing for the internal temperature evolution of LIBs, which is crucial for informing the design of next-generation intelligent thermal management systems.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127394"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A critical review of remaining useful life prediction for water Electrolyzers: From degradation mechanisms to prognostic models 对水电解槽剩余使用寿命预测的评述：从降解机制到预测模型

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2025.127303

Xi Jiang , Zhongkai Zhao , Wenjun Zhou , Linjie Meng , Danfeng Hu , Zixuan Shangguan , Min Liu , Jian Zuo , Cunman Zhang , Liming Jin

With the increasing integration of electrolyzers in energy systems, operational durability has become a key concern. Achieving durability optimization requires effective performance indicators that capture degradation behavior and enable real-time monitoring. Remaining useful life (RUL) prediction, widely applied in batteries and fuel cells, offers a dynamic assessment of equipment health via degradation modeling. However, research on electrolyzer degradation mechanisms is still limited, and RUL lacks standardization and practical applicability in this context. For electrolyzers, DOE targets stack lifetimes of 40,000–80,000 h with degradation rates below 0.2 %/1000 h. However, in practice, the end of life is still commonly defined using an empirical criterion, typically defined as a 5–10 % increase in operating voltage relative to its initial value at a given current density. This study reviews degradation factors, synthesizes the current status of RUL research in electrochemical systems, and proposes a life prediction framework including data preprocessing, modeling, and evaluation. It also analyzes its applicability under electrolyzer conditions and outlines key challenges and future directions toward a reliable and interpretable RUL prediction system.

随着电解槽在能源系统中的集成程度越来越高，运行耐久性已成为一个关键问题。实现耐久性优化需要有效的性能指标来捕捉退化行为并实现实时监控。剩余使用寿命（RUL）预测广泛应用于电池和燃料电池，它通过退化建模提供了对设备健康状况的动态评估。然而，对电解槽降解机理的研究仍然有限，RUL在这一背景下缺乏标准化和实用性。对于电解槽，DOE的目标是电池寿命达到40,000-80,000 h，降解率低于0.2% /1000 h。然而，在实践中，寿命的结束通常仍然使用经验标准来定义，通常定义为在给定电流密度下，工作电压相对于其初始值增加5 - 10%。本文综述了降解因素，综合了电化学系统RUL研究现状，提出了包括数据预处理、建模和评估在内的寿命预测框架。本文还分析了其在电解槽条件下的适用性，并概述了实现可靠和可解释的RUL预测系统的关键挑战和未来方向。

{"title":"A critical review of remaining useful life prediction for water Electrolyzers: From degradation mechanisms to prognostic models","authors":"Xi Jiang , Zhongkai Zhao , Wenjun Zhou , Linjie Meng , Danfeng Hu , Zixuan Shangguan , Min Liu , Jian Zuo , Cunman Zhang , Liming Jin","doi":"10.1016/j.apenergy.2025.127303","DOIUrl":"10.1016/j.apenergy.2025.127303","url":null,"abstract":"<div><div>With the increasing integration of electrolyzers in energy systems, operational durability has become a key concern. Achieving durability optimization requires effective performance indicators that capture degradation behavior and enable real-time monitoring. Remaining useful life (RUL) prediction, widely applied in batteries and fuel cells, offers a dynamic assessment of equipment health via degradation modeling. However, research on electrolyzer degradation mechanisms is still limited, and RUL lacks standardization and practical applicability in this context. For electrolyzers, DOE targets stack lifetimes of 40,000–80,000 h with degradation rates below 0.2 %/1000 h. However, in practice, the end of life is still commonly defined using an empirical criterion, typically defined as a 5–10 % increase in operating voltage relative to its initial value at a given current density. This study reviews degradation factors, synthesizes the current status of RUL research in electrochemical systems, and proposes a life prediction framework including data preprocessing, modeling, and evaluation. It also analyzes its applicability under electrolyzer conditions and outlines key challenges and future directions toward a reliable and interpretable RUL prediction system.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127303"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Natural hydrogen techno-economics and valuation 天然氢的技术经济与估值

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2026.127406

Ning Lin , Aliaksei Patonia , Mark Shuster , Martin Lambert , Tongwei Zhang

Natural hydrogen, often referred to as “white” and sometimes “gold”, has recently emerged as a potentially compelling low-carbon energy resource with significant global decarbonization potential. This study presents a techno-economic analysis using a hybrid approach: it models hydrogen production based on test-well data from the Bourakébougou field in Mali—an ideal reference case due to its high purity and shallow depth, allowing for an assessment of economic potential under favorable geological conditions, while applying a capital and operational cost structure bench-marked against the U.S. Gulf Coast. The estimated levelized cost of hydrogen (LCOH) in a scaled production scenario of 1.3 tons/day across ten vertical wells is $6.82/kg, and it could decrease to $2.46/kg as the scale of production increases to sixty wells without diminishing returns of production. The analysis further demonstrates that these costs are highly sensitive to long-term reservoir performance to complement the constant production rate assumption in the base case. A simplified production decline scenario illustrates an increase in the LCOH by nearly 70%. Besides scale optimization, policy incentives can significantly reduce the LCOH to competitive levels close to $1.00/kg compared to conventional hydrogen production. This study further highlights that uncertainties in resource viability from exploration risks and transportation costs greatly affect the commercial viability of natural hydrogen production.

天然氢通常被称为“白色”，有时也被称为“黄金”，最近成为一种潜在的引人注目的低碳能源，具有巨大的全球脱碳潜力。本研究采用混合方法进行了技术经济分析：基于马里bourak bougou油田的测试井数据（由于其高纯度和浅深度，这是一个理想的参考案例）建立了氢气生产模型，允许在有利的地质条件下评估经济潜力，同时应用以美国墨西哥湾为基准的资本和运营成本结构。在10口直井的1.3吨/天的规模生产场景中，氢气（LCOH）的平均成本估计为6.82美元/公斤，当生产规模增加到60口井时，成本可能降至2.46美元/公斤，而生产回报不会减少。分析进一步表明，这些成本对长期油藏动态高度敏感，以补充基本情况下的恒定产量假设。在简化的产量下降情景中，LCOH增加了近70%。除了规模优化之外，与传统制氢相比，政策激励措施可以将LCOH显著降低到接近1.00美元/公斤的竞争水平。该研究进一步强调，勘探风险和运输成本对资源可行性的不确定性极大地影响了天然制氢的商业可行性。

{"title":"Natural hydrogen techno-economics and valuation","authors":"Ning Lin , Aliaksei Patonia , Mark Shuster , Martin Lambert , Tongwei Zhang","doi":"10.1016/j.apenergy.2026.127406","DOIUrl":"10.1016/j.apenergy.2026.127406","url":null,"abstract":"<div><div>Natural hydrogen, often referred to as “white” and sometimes “gold”, has recently emerged as a potentially compelling low-carbon energy resource with significant global decarbonization potential. This study presents a techno-economic analysis using a hybrid approach: it models hydrogen production based on test-well data from the Bourakébougou field in Mali—an ideal reference case due to its high purity and shallow depth, allowing for an assessment of economic potential under favorable geological conditions, while applying a capital and operational cost structure bench-marked against the U.S. Gulf Coast. The estimated levelized cost of hydrogen (LCOH) in a scaled production scenario of 1.3 tons/day across ten vertical wells is $6.82/kg, and it could decrease to $2.46/kg as the scale of production increases to sixty wells without diminishing returns of production. The analysis further demonstrates that these costs are highly sensitive to long-term reservoir performance to complement the constant production rate assumption in the base case. A simplified production decline scenario illustrates an increase in the LCOH by nearly 70%. Besides scale optimization, policy incentives can significantly reduce the LCOH to competitive levels close to $1.00/kg compared to conventional hydrogen production. This study further highlights that uncertainties in resource viability from exploration risks and transportation costs greatly affect the commercial viability of natural hydrogen production.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127406"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975798","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

High-accuracy dynamic model of high-temperature sodium-sulfur stationary battery 高温钠硫固定电池高精度动力学模型

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2026.127414

Milán Attila Sőrés, Bálint Hartmann

The increasing demand for stationary energy storage has put alternative battery chemistries in the spotlight. Sodium-based technologies appear to be a cost-effective and sustainable option, potentially competing with lithium-ion batteries. However, the long-term reliability, essential for stationary storage services, requires an understanding of performance degradation under realistic operating conditions. In this work, we present a physics-based dynamic equivalent circuit model for Sodium-Sulfur batteries, developed to capture the complexity of system behavior and examine the processes contributing to battery aging. In this way, existing battery models completely overlook the dynamic behavior of sodium‑sulfur batteries. The presented model is validated against real measurements taken at the HUN-REN Centre for Energy Research, where a Sodium-Sulfur battery was recently commissioned. By comparing the performance of the model across various power profiles, its ability to capture the complexity of the entire system is evaluated. Overall simulation results are within ± 1% for most scenarios. The presented work was developed under the 2021–2.1.1-EK funding scheme.

对固定能量存储的需求不断增长，使替代电池化学成为人们关注的焦点。钠基技术似乎是一种具有成本效益和可持续性的选择，有可能与锂离子电池竞争。然而，对于固定存储服务至关重要的长期可靠性需要了解实际操作条件下的性能下降。在这项工作中，我们提出了一个基于物理的钠硫电池动态等效电路模型，该模型旨在捕捉系统行为的复杂性，并检查导致电池老化的过程。这样，现有的电池模型完全忽略了钠硫电池的动态行为。所提出的模型是根据HUN-REN能源研究中心的实际测量结果进行验证的，该中心最近委托了一种钠硫电池。通过比较模型在各种功率配置文件中的性能，可以评估其捕获整个系统复杂性的能力。在大多数情况下，总体模拟结果在±1%以内。所介绍的工作是根据2021-2.1.1-EK资助计划进行的。

{"title":"High-accuracy dynamic model of high-temperature sodium-sulfur stationary battery","authors":"Milán Attila Sőrés, Bálint Hartmann","doi":"10.1016/j.apenergy.2026.127414","DOIUrl":"10.1016/j.apenergy.2026.127414","url":null,"abstract":"<div><div>The increasing demand for stationary energy storage has put alternative battery chemistries in the spotlight. Sodium-based technologies appear to be a cost-effective and sustainable option, potentially competing with lithium-ion batteries. However, the long-term reliability, essential for stationary storage services, requires an understanding of performance degradation under realistic operating conditions. In this work, we present a physics-based dynamic equivalent circuit model for Sodium-Sulfur batteries, developed to capture the complexity of system behavior and examine the processes contributing to battery aging. In this way, existing battery models completely overlook the dynamic behavior of sodium‑sulfur batteries. The presented model is validated against real measurements taken at the HUN-REN Centre for Energy Research, where a Sodium-Sulfur battery was recently commissioned. By comparing the performance of the model across various power profiles, its ability to capture the complexity of the entire system is evaluated. Overall simulation results are within ± 1% for most scenarios. The presented work was developed under the 2021–2.1.1-EK funding scheme.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127414"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A reduced-order model for predicting transient performance of air-source heat pumps 空气源热泵暂态性能预测的降阶模型

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2026.127412

Shahzad Yousaf, Syed Mohammad Meesam Raza, Amjid Khan, Craig R. Bradshaw

This paper presents a transient, reduced-order model for variable-speed air-source heat pumps that captures start-up and cycling dynamics using only air-side inputs in heating operation. The model integrates two sub-models: a heating-capacity and a compressor power model identified via symbolic regression using high-fidelity simulations and experimental data. Validation was conducted on a 4-ton variable-speed heat pump tested in twin psychrometric chambers under diverse steady-state and dynamic start-up conditions. The models reproduce steady state operation with a mean absolute percentage error (MAPE) of 2.5% and 1.6% in heating capacity and power consumption, respectively. Dynamic error metrics remain below 4% for both the coefficient of variation of root mean square error (CVRMSE) and normalized mean bias error (NMBE) using only air-side temperatures, indoor fan supply, and compressor speed along with estimates of heat exchanger and zone air thermal mass. Sensitivity analyses confirm robustness to

\pm 20 %

uncertainty in thermal mass assumptions. Unlike traditional approaches that rely on fixed degradation constants, the model explicitly differentiates between cold- and hot-start transients, enabling more accurate representation of start-up behavior. A representative cold-climate (Chicago, Illinois, USA) case study indicates that neglecting transient dynamics can bias seasonal performance and peak demand estimates by up to 4.9%. By enabling fast and accurate heat pump performance predictions, the model bridges the gap between high-fidelity physics-based simulations and the practical needs of building performance modeling while achieving simulation speeds over

10^{6}

times faster than real time.

本文提出了一种变速空气源热泵的瞬态降阶模型，该模型仅使用加热操作中的空气侧输入来捕获启动和循环动力学。该模型集成了两个子模型：加热容量和压缩机功率模型，通过使用高保真仿真和实验数据进行符号回归确定。对一台4吨变速热泵在双湿室中进行了不同稳态和动态启动条件下的验证。模型模拟稳态运行时，供热能力和能耗的平均绝对百分比误差（MAPE）分别为2.5%和1.6%。仅使用空气侧温度、室内风机供气量、压缩机转速以及热交换器和区域空气热质量的估计，均方根误差变异系数（CVRMSE）和归一化平均偏差（NMBE）的动态误差指标均保持在4%以下。灵敏度分析证实了热质量假设对±20%不确定性的稳健性。与依赖固定退化常数的传统方法不同，该模型明确区分了冷启动和热启动瞬态，从而能够更准确地表示启动行为。一个具有代表性的寒冷气候（美国伊利诺斯州芝加哥）案例研究表明，忽略瞬态动态可以使季节性性能和峰值需求估计偏差高达4.9%。通过实现快速准确的热泵性能预测，该模型弥合了高保真物理模拟与建筑性能建模实际需求之间的差距，同时实现了比实时快106倍以上的仿真速度。

{"title":"A reduced-order model for predicting transient performance of air-source heat pumps","authors":"Shahzad Yousaf, Syed Mohammad Meesam Raza, Amjid Khan, Craig R. Bradshaw","doi":"10.1016/j.apenergy.2026.127412","DOIUrl":"10.1016/j.apenergy.2026.127412","url":null,"abstract":"<div><div>This paper presents a transient, reduced-order model for variable-speed air-source heat pumps that captures start-up and cycling dynamics using only air-side inputs in heating operation. The model integrates two sub-models: a heating-capacity and a compressor power model identified via symbolic regression using high-fidelity simulations and experimental data. Validation was conducted on a 4-ton variable-speed heat pump tested in twin psychrometric chambers under diverse steady-state and dynamic start-up conditions. The models reproduce steady state operation with a mean absolute percentage error (MAPE) of 2.5% and 1.6% in heating capacity and power consumption, respectively. Dynamic error metrics remain below 4% for both the coefficient of variation of root mean square error (CVRMSE) and normalized mean bias error (NMBE) using only air-side temperatures, indoor fan supply, and compressor speed along with estimates of heat exchanger and zone air thermal mass. Sensitivity analyses confirm robustness to <span><math><mo>±</mo><mn>20</mn><mi>%</mi></math></span> uncertainty in thermal mass assumptions. Unlike traditional approaches that rely on fixed degradation constants, the model explicitly differentiates between cold- and hot-start transients, enabling more accurate representation of start-up behavior. A representative cold-climate (Chicago, Illinois, USA) case study indicates that neglecting transient dynamics can bias seasonal performance and peak demand estimates by up to 4.9%. By enabling fast and accurate heat pump performance predictions, the model bridges the gap between high-fidelity physics-based simulations and the practical needs of building performance modeling while achieving simulation speeds over <span><math><msup><mn>10</mn><mrow><mn>6</mn></mrow></msup></math></span> times faster than real time.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127412"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975796","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Electrode microstructure-driven multi-scale feature fusion framework based on microscopic imaging technique for health estimate of lithium-ion batteries 基于显微成像技术的电极微结构驱动多尺度特征融合框架用于锂离子电池健康评估

IF 11 1区工程技术 Q1 ENERGY & FUELS

Applied Energy

Pub Date : 2026-01-16 DOI: 10.1016/j.apenergy.2026.127398

Hongmin Jiang , Qiangxiang Zhai , Nengbing Long , Qiaoling Kang , Xianhe Meng , Mingjiong Zhou , Lijing Yan , Tingli Ma

Accurately estimating the health of lithium-ion batteries is crucial for ensuring their long-term reliability and stable performance. Traditional estimating methods, primarily based on electrical and thermal signal analyses, do not account for microstructural changes predominantly characteristic of capacity loss and performance degradation. Therefore, this study introduces a novel health estimation framework based on scanning electron microscopy integrated with cross-modal fusion to extract morphological and high-dimensional structural features. Pyramid-attention and region-adaptive fusion strategies are developed to combine information obtained at different magnifications and from different electrode regions, thereby balancing detailed microstructural characteristics and overarching degradation trends. Experimental results indicate that this hybrid feature fusion approach, incorporating data augmentation and synthetic sample generation, improves estimation accuracy and model generalization. The pyramid-attention mechanism optimizes multi-magnification information integration, and region-adaptive weighting enhances the degradation assessment stability. The proposed model outperforms single-modality approaches and achieves a root mean squared error of 1.25% under optimal conditions, with the enhancements contributing to an error reduction of up to 65.7% compared with that under non-enhanced datasets. The proposed method leverages microstructural insights to offer a scalable solution for battery health monitoring and material development and a deeper understanding of degradation mechanisms and battery performance improvements.

准确评估锂离子电池的健康状况对于确保其长期可靠性和稳定性能至关重要。传统的评估方法主要基于电和热信号分析，不能考虑以容量损失和性能退化为主要特征的微观结构变化。因此，本研究引入了一种基于扫描电镜结合跨模态融合的新型健康估计框架，以提取形态学和高维结构特征。金字塔关注和区域自适应融合策略被开发出来，以结合在不同的放大倍率和不同的电极区域获得的信息，从而平衡详细的微观结构特征和总体退化趋势。实验结果表明，结合数据增强和合成样本生成的混合特征融合方法提高了估计精度和模型泛化能力。金字塔关注机制优化了多倍信息整合，区域自适应加权增强了退化评价的稳定性。该模型优于单模态方法，在最优条件下实现了1.25%的均方根误差，与未增强数据集相比，增强后的模型误差降低了65.7%。所提出的方法利用微观结构的见解，为电池健康监测和材料开发提供可扩展的解决方案，并更深入地了解降解机制和电池性能改进。

{"title":"Electrode microstructure-driven multi-scale feature fusion framework based on microscopic imaging technique for health estimate of lithium-ion batteries","authors":"Hongmin Jiang , Qiangxiang Zhai , Nengbing Long , Qiaoling Kang , Xianhe Meng , Mingjiong Zhou , Lijing Yan , Tingli Ma","doi":"10.1016/j.apenergy.2026.127398","DOIUrl":"10.1016/j.apenergy.2026.127398","url":null,"abstract":"<div><div>Accurately estimating the health of lithium-ion batteries is crucial for ensuring their long-term reliability and stable performance. Traditional estimating methods, primarily based on electrical and thermal signal analyses, do not account for microstructural changes predominantly characteristic of capacity loss and performance degradation. Therefore, this study introduces a novel health estimation framework based on scanning electron microscopy integrated with cross-modal fusion to extract morphological and high-dimensional structural features. Pyramid-attention and region-adaptive fusion strategies are developed to combine information obtained at different magnifications and from different electrode regions, thereby balancing detailed microstructural characteristics and overarching degradation trends. Experimental results indicate that this hybrid feature fusion approach, incorporating data augmentation and synthetic sample generation, improves estimation accuracy and model generalization. The pyramid-attention mechanism optimizes multi-magnification information integration, and region-adaptive weighting enhances the degradation assessment stability. The proposed model outperforms single-modality approaches and achieves a root mean squared error of 1.25% under optimal conditions, with the enhancements contributing to an error reduction of up to 65.7% compared with that under non-enhanced datasets. The proposed method leverages microstructural insights to offer a scalable solution for battery health monitoring and material development and a deeper understanding of degradation mechanisms and battery performance improvements.</div></div>","PeriodicalId":246,"journal":{"name":"Applied Energy","volume":"408 ","pages":"Article 127398"},"PeriodicalIF":11.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145975795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0