Energy and AI最新文献_第2页

A World Model framework with digital twin for scheduling demand-side resources under systemic complexity 系统复杂性下需求侧资源调度的数字孪生世界模型框架

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2025.100660

Xing He , Yikang Bu , Guoquan Yuan , Junjie Yin , Zhuangyan Zhang , Qian Ai , Caiming Qiu

Modern power grids face profound challenges in scheduling massive, heterogeneous demand-side resources (DSRs), whose collective behaviors often lead to systemic unpredictability and scheduling inaccuracies. Traditional methods, often based on simplified models, struggle to manage this emergent complexity. To address this gap, this paper introduces MetaGrid, a novel digital-twin-enhanced World Model framework designed for proactive and prescient DSR scheduling. The MetaGrid architecture is composed of four integral, closed-loop building blocks: a General Simulator for multi-path deduction, a Situational Perceiver for holistic cognition, an Intelligent Decision-Maker for autonomous optimization, and a Unified Verifier for ensuring trustworthy iteration. By integrating principles from complexity science with data-intensive machine learning, MetaGrid creates a high-fidelity metaverse to model and manage DSR ecosystems. The framework’s core capabilities are demonstrated through an energy storage system scheduling case, where an intelligent agent, guided by the World Model, learns to autonomously balance real-time electricity costs against physical battery degradation constraints. This preliminary validation showcases MetaGrid’s potential as a powerful tool for navigating the complexities of future energy systems, transforming scheduling from a reactive control problem into a process of continuous, adaptive learning.

现代电网在调度海量异构需求侧资源（dsr）方面面临着深刻的挑战，这些资源的集体行为往往导致系统的不可预测性和调度不准确性。传统的方法，通常基于简化的模型，难以管理这种突发的复杂性。为了解决这一差距，本文介绍了MetaGrid，这是一种新型的数字孪生增强世界模型框架，旨在实现前瞻性和先见之明的DSR调度。MetaGrid架构由四个完整的闭环构建块组成：一个用于多路径演绎的通用模拟器，一个用于整体认知的情境感知器，一个用于自主优化的智能决策者，以及一个用于确保可信迭代的统一验证器。通过将复杂性科学原理与数据密集型机器学习相结合，MetaGrid创建了一个高保真的元宇宙来建模和管理DSR生态系统。该框架的核心功能通过一个储能系统调度案例来展示，其中智能代理在世界模型的指导下，学习自主平衡实时电力成本和物理电池退化约束。这一初步验证显示了MetaGrid作为导航未来能源系统复杂性的强大工具的潜力，将调度从被动控制问题转变为持续、自适应学习的过程。

{"title":"A World Model framework with digital twin for scheduling demand-side resources under systemic complexity","authors":"Xing He , Yikang Bu , Guoquan Yuan , Junjie Yin , Zhuangyan Zhang , Qian Ai , Caiming Qiu","doi":"10.1016/j.egyai.2025.100660","DOIUrl":"10.1016/j.egyai.2025.100660","url":null,"abstract":"<div><div>Modern power grids face profound challenges in scheduling massive, heterogeneous demand-side resources (DSRs), whose collective behaviors often lead to systemic unpredictability and scheduling inaccuracies. Traditional methods, often based on simplified models, struggle to manage this emergent complexity. To address this gap, this paper introduces MetaGrid, a novel digital-twin-enhanced World Model framework designed for proactive and prescient DSR scheduling. The MetaGrid architecture is composed of four integral, closed-loop building blocks: a General Simulator for multi-path deduction, a Situational Perceiver for holistic cognition, an Intelligent Decision-Maker for autonomous optimization, and a Unified Verifier for ensuring trustworthy iteration. By integrating principles from complexity science with data-intensive machine learning, MetaGrid creates a high-fidelity metaverse to model and manage DSR ecosystems. The framework’s core capabilities are demonstrated through an energy storage system scheduling case, where an intelligent agent, guided by the World Model, learns to autonomously balance real-time electricity costs against physical battery degradation constraints. This preliminary validation showcases MetaGrid’s potential as a powerful tool for navigating the complexities of future energy systems, transforming scheduling from a reactive control problem into a process of continuous, adaptive learning.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100660"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926796","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

Practicality-enhanced behind-the-meter PV power generation disaggregation based on synchronization and transferability fused LSTM framework 实用性增强的基于同步和可转移性融合LSTM框架的光伏发电分解

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100675

Chengye Zhang , Huan Long , Zijun Zhang , Jinde Cao

To facilitate the operation of distribution networks with a large scale of household photovoltaic systems integrated, the availability of community-level behind-the-meter (BTM) PV power generation is crucial. Yet, due to the scarcity of smart meters installed, it is challenging to obtain such information via directly aggregating measured power outputs of individual PV systems, and an effective estimation method needs to be developed. Considering the similarity between household-level and community-level data within the same geographical area, this paper develops a synchronization and model-transfer fused LSTM framework (SAM-LSTM). The core technical contribution lies in the development of the Synchronized Long Short-Term Memory (Syn-LSTM), which separately models the synchronized factors and disaggregated BTM data to capture more generalized representations. The learned household-level representations are then transferred to the community-level. Finally, by explicitly leveraging the complementarity between PV generation and consumption, a dual time-series modeling architecture is developed to refine the initial community-level PV power generation estimates, thereby alleviating potential biases introduced during the model-transfer process. Extensive computational studies are conducted to demonstrate the effectiveness of SAM-LSTM in community-level BTM PV power generation disaggregation in real data from Hebei, China. Compared with the best-performing benchmarks, SAM-LSTM achieves up to 56% lower MSE, significantly demonstrating its strong generalization and robustness capabilities.

为了促进大规模家庭光伏系统集成配电网络的运行，社区级电表后（BTM）光伏发电的可用性至关重要。然而，由于安装的智能电表数量稀少，通过直接汇总单个光伏系统的实测功率输出来获取这些信息是具有挑战性的，需要开发一种有效的估算方法。考虑到同一地理区域内家庭级和社区级数据的相似性，本文提出了一种同步和模型转移融合的LSTM框架（SAM-LSTM）。其核心技术贡献在于同步长短期记忆（Syn-LSTM）的发展，它分别对同步因素和分解的BTM数据进行建模，以获取更广义的表征。学习到的家庭层面的表征然后被转移到社区层面。最后，通过明确利用光伏发电和消费之间的互补性，开发了双时间序列建模架构，以改进初始社区级光伏发电估计，从而减轻模型迁移过程中引入的潜在偏差。通过大量的计算研究，验证了SAM-LSTM在社区BTM光伏发电分类中的有效性。与性能最好的基准测试相比，SAM-LSTM的MSE降低了56%，显著证明了其强大的泛化和鲁棒性。

{"title":"Practicality-enhanced behind-the-meter PV power generation disaggregation based on synchronization and transferability fused LSTM framework","authors":"Chengye Zhang , Huan Long , Zijun Zhang , Jinde Cao","doi":"10.1016/j.egyai.2026.100675","DOIUrl":"10.1016/j.egyai.2026.100675","url":null,"abstract":"<div><div>To facilitate the operation of distribution networks with a large scale of household photovoltaic systems integrated, the availability of community-level behind-the-meter (BTM) PV power generation is crucial. Yet, due to the scarcity of smart meters installed, it is challenging to obtain such information via directly aggregating measured power outputs of individual PV systems, and an effective estimation method needs to be developed. Considering the similarity between household-level and community-level data within the same geographical area, this paper develops a synchronization and model-transfer fused LSTM framework (SAM-LSTM). The core technical contribution lies in the development of the Synchronized Long Short-Term Memory (Syn-LSTM), which separately models the synchronized factors and disaggregated BTM data to capture more generalized representations. The learned household-level representations are then transferred to the community-level. Finally, by explicitly leveraging the complementarity between PV generation and consumption, a dual time-series modeling architecture is developed to refine the initial community-level PV power generation estimates, thereby alleviating potential biases introduced during the model-transfer process. Extensive computational studies are conducted to demonstrate the effectiveness of SAM-LSTM in community-level BTM PV power generation disaggregation in real data from Hebei, China. Compared with the best-performing benchmarks, SAM-LSTM achieves up to 56% lower MSE, significantly demonstrating its strong generalization and robustness capabilities.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100675"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977684","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

A transformer-LSTM network enhanced by EEMD for ultra-short-term wind power forecasting 基于EEMD的超短期风电预测变压器- lstm网络

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100682

YongSheng Wang, Fan Yang, YongSheng Qi, GuangChen Liu, JiaJing Gao, XueHui Wang, ZhenChao Wang

This study aims to improve the dispatch safety and economic efficiency of grid-connected wind power systems by addressing the limitations of traditional ultra-short-term forecasting methods, particularly their inadequate extraction of multi-scale features and limited forecasting accuracy. A short-term wind power forecasting model that integrates signal decomposition with deep learning is proposed. The model first applies Ensemble Empirical Mode Decomposition (EEMD) to the raw wind power data to reduce non-stationarity and extract multi-scale features. A lightweight Transformer attention mechanism is then employed to model global dependencies, and Long Short-Term Memory (LSTM) networks are incorporated to capture the temporal dynamics of the sequence. The final power forecasting is generated through a fully connected layer. Finally, the Alpha Evolution (AE) algorithm is employed to optimize the model's hyperparameters. Experiments on multiple datasets show that the proposed model outperforms traditional machine learning and deep learning approaches across various evaluation metrics. It achieves higher fitting accuracy, confirming its effectiveness and robustness in multi-scale feature extraction and wind power forecasting.

本研究旨在解决传统超短期预测方法对多尺度特征提取不足、预测精度有限的局限性，提高并网风电系统的调度安全性和经济性。提出了一种将信号分解与深度学习相结合的短期风电预测模型。该模型首先将集成经验模态分解（Ensemble Empirical Mode Decomposition， EEMD）应用于原始风电数据，以减少非平稳性并提取多尺度特征。然后使用轻量级的Transformer注意机制对全局依赖性进行建模，并结合长短期记忆（LSTM）网络来捕获序列的时间动态。最终的功率预测是通过一个全连通层生成的。最后，采用Alpha Evolution （AE）算法对模型的超参数进行优化。在多个数据集上的实验表明，该模型在各种评估指标上优于传统的机器学习和深度学习方法。该方法取得了较高的拟合精度，验证了其在多尺度特征提取和风电预测中的有效性和鲁棒性。

{"title":"A transformer-LSTM network enhanced by EEMD for ultra-short-term wind power forecasting","authors":"YongSheng Wang, Fan Yang, YongSheng Qi, GuangChen Liu, JiaJing Gao, XueHui Wang, ZhenChao Wang","doi":"10.1016/j.egyai.2026.100682","DOIUrl":"10.1016/j.egyai.2026.100682","url":null,"abstract":"<div><div>This study aims to improve the dispatch safety and economic efficiency of grid-connected wind power systems by addressing the limitations of traditional ultra-short-term forecasting methods, particularly their inadequate extraction of multi-scale features and limited forecasting accuracy. A short-term wind power forecasting model that integrates signal decomposition with deep learning is proposed. The model first applies Ensemble Empirical Mode Decomposition (EEMD) to the raw wind power data to reduce non-stationarity and extract multi-scale features. A lightweight Transformer attention mechanism is then employed to model global dependencies, and Long Short-Term Memory (LSTM) networks are incorporated to capture the temporal dynamics of the sequence. The final power forecasting is generated through a fully connected layer. Finally, the Alpha Evolution (AE) algorithm is employed to optimize the model's hyperparameters. Experiments on multiple datasets show that the proposed model outperforms traditional machine learning and deep learning approaches across various evaluation metrics. It achieves higher fitting accuracy, confirming its effectiveness and robustness in multi-scale feature extraction and wind power forecasting.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100682"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926794","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

A bi-level advanced control framework for large-scale control of buildings with system-level impact 具有系统级影响的建筑物大规模控制的两级高级控制框架

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100678

Dylan Wald , Olga Doronina , Kathryn Johnson , Ryan King , Michael Sinner , Kevin Griffin , Rohit Chintala , Deepthi Vaidhynathan , Jibonananda Sanyal , Marc Day

Increased electricity consumption combined with new forms of generation is testing the reliability of our grid infrastructure. This work describes a method to improve the reliability of the grid through large-scale advanced building control. This paper develops a bi-level distributed control framework to shift the load of 153 buildings to achieve a system-level objective of tracking a power reference signal. This bi-level control is based on the previously-developed ANPV-MPC, a predictive controller that uses a Bayesian neural network to generate an accurate control model and adapt to changing conditions over time. By shifting the building electricity demand to better match the available power, the grid system supplying the buildings is more reliable as evidenced by the analysis of node voltages across an IEEE 13-bus distribution system. The proposed bi-level control framework tracks the system-level power reference with enough accuracy to regulate node voltages across the IEEE 13-bus distribution system within ANSI limits of

\pm

5%. Additionally, the adaptive nature of ANPV-MPC allows each building across the system to adapt to changing conditions, further amplifying the system-level reliability.

不断增加的电力消耗与新型发电方式相结合，正在考验我们电网基础设施的可靠性。本文介绍了一种通过大型先进建筑控制来提高电网可靠性的方法。本文开发了一种双层分布式控制框架，用于转移153个建筑物的负载，以实现跟踪电源参考信号的系统级目标。这种双级控制基于先前开发的ANPV-MPC，这是一种预测控制器，使用贝叶斯神经网络生成精确的控制模型，并随着时间的推移适应不断变化的条件。通过对IEEE 13总线配电系统节点电压的分析，可以证明通过改变建筑物电力需求以更好地匹配可用电力，为建筑物供电的电网系统更加可靠。所提出的双电平控制框架以足够的精度跟踪系统级功率参考，以在±5%的ANSI限制内调节IEEE 13总线配电系统的节点电压。此外，ANPV-MPC的自适应特性允许整个系统中的每个建筑适应不断变化的条件，进一步增强了系统级的可靠性。

{"title":"A bi-level advanced control framework for large-scale control of buildings with system-level impact","authors":"Dylan Wald , Olga Doronina , Kathryn Johnson , Ryan King , Michael Sinner , Kevin Griffin , Rohit Chintala , Deepthi Vaidhynathan , Jibonananda Sanyal , Marc Day","doi":"10.1016/j.egyai.2026.100678","DOIUrl":"10.1016/j.egyai.2026.100678","url":null,"abstract":"<div><div>Increased electricity consumption combined with new forms of generation is testing the reliability of our grid infrastructure. This work describes a method to improve the reliability of the grid through large-scale advanced building control. This paper develops a bi-level distributed control framework to shift the load of 153 buildings to achieve a system-level objective of tracking a power reference signal. This bi-level control is based on the previously-developed ANPV-MPC, a predictive controller that uses a Bayesian neural network to generate an accurate control model and adapt to changing conditions over time. By shifting the building electricity demand to better match the available power, the grid system supplying the buildings is more reliable as evidenced by the analysis of node voltages across an IEEE 13-bus distribution system. The proposed bi-level control framework tracks the system-level power reference with enough accuracy to regulate node voltages across the IEEE 13-bus distribution system within ANSI limits of <span><math><mo>±</mo></math></span>5%. Additionally, the adaptive nature of ANPV-MPC allows each building across the system to adapt to changing conditions, further amplifying the system-level reliability.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100678"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022844","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

Spectrogram-driven unsupervised autoencoder with isolation forest and one-class SVM for lab-scale wind turbine blade fault detection 基于隔离森林和一类支持向量机的谱图驱动无监督自编码器用于风力发电机叶片故障检测

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100681

Waqar Ali , Idriss El-Thalji , Knut Erik Teigen Giljarhus , Andreas Delimitis

Wind turbine blades are critical components, and their structural integrity is essential for uninterrupted operation and minimizing downtime. Although various methods are used to monitor the health of wind turbine blades, several research challenges persist, such as the reliance on manual feature engineering and the limited availability of large amounts of labeled data. In this study, a novel approach is proposed that will overcome the limitations of manual feature extraction and label data challenges. In the proposed work, time series vibration signals from the blade are first converted into spectrograms and passed through a CNN-based autoencoder that is trained solely on healthy data to learn a compact latent representation. Anomalies are then flagged in three complementary ways: (i) by thresholding the autoencoder’s reconstruction error, (ii) by applying an Isolation Forest to the latent features, and (iii) by evaluating the same features with a One-Class SVM. The outputs of these detectors are subsequently benchmarked, providing a systematic comparison of their ability to discriminate between vibration-induced faults, such as cracks and erosion, and normal operation on a controlled test-rig dataset, the autoencoder achieves 97.2 % accuracy, outperforming the Isolation Forest and One-Class SVM by 8%–27%. These results demonstrate that zero-label, deep-feature pipelines can deliver reliable and scalable blade-fault detection, paving the way for more cost-effective predictive maintenance in wind farms.

风力涡轮机叶片是关键部件，其结构完整性对于不间断运行和减少停机时间至关重要。尽管有多种方法用于监测风力涡轮机叶片的健康状况，但仍存在一些研究挑战，例如依赖于手动特征工程和大量标记数据的有限可用性。在这项研究中，提出了一种新的方法来克服人工特征提取的局限性和标记数据的挑战。在本文中，来自叶片的时间序列振动信号首先被转换成频谱图，并通过基于cnn的自编码器，该自编码器仅在健康数据上进行训练，以学习紧凑的潜在表示。然后以三种互补的方式标记异常：(i)通过对自编码器的重建误差设置阈值，（ii）通过对潜在特征应用隔离森林，以及（iii）通过使用一类支持向量机评估相同的特征。随后对这些检测器的输出进行基准测试，对它们区分振动引起的故障（如裂纹和侵蚀）和受控试验台数据集的正常运行的能力进行系统比较，自动编码器达到97.2%的准确率，优于隔离森林和一类支持向量机8%-27%。这些结果表明，零标签、深度特征管道可以提供可靠的、可扩展的叶片故障检测，为风电场更具成本效益的预测性维护铺平了道路。

{"title":"Spectrogram-driven unsupervised autoencoder with isolation forest and one-class SVM for lab-scale wind turbine blade fault detection","authors":"Waqar Ali , Idriss El-Thalji , Knut Erik Teigen Giljarhus , Andreas Delimitis","doi":"10.1016/j.egyai.2026.100681","DOIUrl":"10.1016/j.egyai.2026.100681","url":null,"abstract":"<div><div>Wind turbine blades are critical components, and their structural integrity is essential for uninterrupted operation and minimizing downtime. Although various methods are used to monitor the health of wind turbine blades, several research challenges persist, such as the reliance on manual feature engineering and the limited availability of large amounts of labeled data. In this study, a novel approach is proposed that will overcome the limitations of manual feature extraction and label data challenges. In the proposed work, time series vibration signals from the blade are first converted into spectrograms and passed through a CNN-based autoencoder that is trained solely on healthy data to learn a compact latent representation. Anomalies are then flagged in three complementary ways: (i) by thresholding the autoencoder’s reconstruction error, (ii) by applying an Isolation Forest to the latent features, and (iii) by evaluating the same features with a One-Class SVM. The outputs of these detectors are subsequently benchmarked, providing a systematic comparison of their ability to discriminate between vibration-induced faults, such as cracks and erosion, and normal operation on a controlled test-rig dataset, the autoencoder achieves 97.2 % accuracy, outperforming the Isolation Forest and One-Class SVM by 8%–27%. These results demonstrate that zero-label, deep-feature pipelines can deliver reliable and scalable blade-fault detection, paving the way for more cost-effective predictive maintenance in wind farms.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100681"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926793","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

Photovoltaic Knowledge-Informed Neural Network (PKINN): Interpretable power prediction model under Fluctuating Environmental Conditions 光伏知识知情神经网络（PKINN）：波动环境下的可解释功率预测模型

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100683

Jialong Pei , Jieming Ma , Ka Lok Man , Martin Gairing

Fluctuating Environmental Conditions (FECs) are a critical barrier to accurate photovoltaic (PV) power forecasting. Existing models often fail to capture abrupt and stochastic fluctuations, leading to reduced forecasting reliability. To address this challenge, this study proposes an interpretable Photovoltaic Knowledge-Informed Neural Network (PKINN). The framework incorporates a Quadratic Explicit Model (QEM) to derive explicit expressions of PV power and transparently capture abrupt variations, while a Fluctuation Allocation Mechanism (FAM) employs a fluctuation sensitivity coefficient to quantify fluctuation intensity and allocate input data to specialized prediction branches. The proposed PKINN framework enables adaptive learning across diverse FECs and enhances forecasting performance. Experimental evaluations on two types of PV modules demonstrate that PKINN reduces the root mean square error by at least 8.73% compared with state-of-the-art models across diverse FECs.

波动环境条件（FECs）是光伏发电（PV）准确预测的关键障碍。现有的模型往往不能捕捉到突然的和随机的波动，导致预测的可靠性降低。为了解决这一挑战，本研究提出了一种可解释的光伏知识知情神经网络（PKINN）。该框架采用二次显式模型（QEM）来推导光伏发电的显式表达式，并透明地捕捉突变变化，而波动分配机制（FAM）采用波动敏感系数来量化波动强度，并将输入数据分配给专门的预测分支。提出的PKINN框架可以实现跨不同fec的自适应学习，并提高预测性能。对两种类型光伏组件的实验评估表明，与不同fec的最先进模型相比，PKINN将均方根误差降低了至少8.73%。

引用次数: 0

TRACE: Time series representation learning with contrastive embeddings for anomaly detection in photovoltaic systems TRACE：光伏系统异常检测的对比嵌入时间序列表示学习

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2025.100670

Chandana Priya Nivarthi, Zhixin Huang, Christian Gruhl, Bernhard Sick

Reliable anomaly detection in photovoltaic (PV) inverters is critical for ensuring operational efficiency and reducing maintenance costs in renewable energy systems. We introduce TRACE (Time series Representation learning with Autoencoder-based Contrastive Embeddings), a self-supervised contrastive learning framework for multivariate time series anomaly detection in PV systems. TRACE employs a two-stage architecture: autoencoder-based representation learning with interchangeable backbones followed by contrastive training through a Siamese network. The framework generates semantically coherent augmentations by perturbing autoencoder reconstructions and applies three negative mining strategies to create challenging contrastive pairs. Comprehensive experiments on a real-world PV inverter dataset and two industrial benchmarks demonstrate TRACE’s superiority. Autoencoder-based augmentations deliver a 21.3% relative improvement in mean F1 (0.616 vs. 0.508) over traditional perturbation methods, with TransformerAE emerging as the optimal backbone architecture. While negative sampling strategies show dataset-specific advantages, their impact remains secondary to encoder capacity. TRACE with TransformerAE and reconstruction-error negatives consistently outperforms fourteen state-of-the-art time series anomaly detection methods, achieving highest F1 scores on all the three datasets while maintaining exceptional precision up to 0.99. Visualization analysis confirms TRACE’s capacity for early fault detection up to three days before failure and interpretable embedding separation. The framework addresses the fundamental challenge of label scarcity in industrial monitoring through self-supervised learning, providing a practical and transparent solution for predictive maintenance in PV systems and broader industrial applications.

在可再生能源系统中，可靠的光伏逆变器异常检测对于确保运行效率和降低维护成本至关重要。我们介绍了TRACE（基于自编码器的对比嵌入的时间序列表示学习），这是一个用于光伏系统中多变量时间序列异常检测的自监督对比学习框架。TRACE采用两阶段架构：基于自动编码器的表示学习，具有可互换的主干，然后通过Siamese网络进行对比训练。该框架通过扰动自编码器重构生成语义连贯增强，并应用三种负挖掘策略来创建具有挑战性的对比对。在实际光伏逆变器数据集和两个工业基准上的综合实验证明了TRACE的优势。与传统的扰动方法相比，基于自编码器的增强方法平均F1相对提高了21.3% (0.616 vs. 0.508)，其中TransformerAE成为了最佳的骨干架构。虽然负采样策略显示了数据集特定的优势，但它们的影响仍然次于编码器容量。具有TransformerAE和重建误差底片的TRACE始终优于14种最先进的时间序列异常检测方法，在所有三个数据集上获得最高的F1分数，同时保持高达0.99的卓越精度。可视化分析证实了TRACE的早期故障检测能力，可在故障发生前三天进行检测，并可解释嵌入分离。该框架通过自我监督学习解决了工业监控中标签稀缺的基本挑战，为光伏系统和更广泛的工业应用的预测性维护提供了实用和透明的解决方案。

{"title":"TRACE: Time series representation learning with contrastive embeddings for anomaly detection in photovoltaic systems","authors":"Chandana Priya Nivarthi, Zhixin Huang, Christian Gruhl, Bernhard Sick","doi":"10.1016/j.egyai.2025.100670","DOIUrl":"10.1016/j.egyai.2025.100670","url":null,"abstract":"<div><div>Reliable anomaly detection in photovoltaic (PV) inverters is critical for ensuring operational efficiency and reducing maintenance costs in renewable energy systems. We introduce TRACE (<strong>T</strong>ime series <strong>R</strong>epresentation learning with <strong>A</strong>utoencoder-based <strong>C</strong>ontrastive <strong>E</strong>mbeddings), a self-supervised contrastive learning framework for multivariate time series anomaly detection in PV systems. TRACE employs a two-stage architecture: autoencoder-based representation learning with interchangeable backbones followed by contrastive training through a Siamese network. The framework generates semantically coherent augmentations by perturbing autoencoder reconstructions and applies three negative mining strategies to create challenging contrastive pairs. Comprehensive experiments on a real-world PV inverter dataset and two industrial benchmarks demonstrate TRACE’s superiority. Autoencoder-based augmentations deliver a 21.3% relative improvement in mean F1 (0.616 vs. 0.508) over traditional perturbation methods, with TransformerAE emerging as the optimal backbone architecture. While negative sampling strategies show dataset-specific advantages, their impact remains secondary to encoder capacity. TRACE with TransformerAE and reconstruction-error negatives consistently outperforms fourteen state-of-the-art time series anomaly detection methods, achieving highest F1 scores on all the three datasets while maintaining exceptional precision up to 0.99. Visualization analysis confirms TRACE’s capacity for early fault detection up to three days before failure and interpretable embedding separation. The framework addresses the fundamental challenge of label scarcity in industrial monitoring through self-supervised learning, providing a practical and transparent solution for predictive maintenance in PV systems and broader industrial applications.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100670"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926792","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

Assessing long-term electricity market design for ambitious decarbonization targets using multi-agent reinforcement learning 使用多智能体强化学习评估雄心勃勃的脱碳目标的长期电力市场设计

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2025.100665

Javier Gonzalez-Ruiz , Carlos Rodriguez-Pardo , Iacopo Savelli , Alice Di Bella , Massimo Tavoni

Electricity systems are key to transforming today’s society into a carbon-free economy. Long-term electricity market mechanisms, including auctions, support schemes, and other policy instruments, are critical in shaping the electricity generation mix. In light of the need for more advanced tools to support policymakers and other stakeholders in designing, testing, and evaluating long-term markets, this work presents a multi-agent reinforcement learning model capable of capturing the key features of decarbonizing energy systems. Profit-maximizing generation companies make investment decisions in the wholesale electricity market, responding to system needs, competitive dynamics, and policy signals. The model employs independent proximal policy optimization, which was selected for suitability to the decentralized and competitive environment. Nevertheless, given the inherent challenges of independent learning in multi-agent settings, an extensive hyperparameter search ensures that decentralized training yields market outcomes consistent with competitive behavior. The model is applied to a stylized version of the Italian electricity system and tested under varying levels of competition, market designs, and policy scenarios. Results highlight the critical role of market design for decarbonizing the electricity sector and avoiding price volatility. The proposed framework allows assessing long-term electricity markets in which multiple policy and market mechanisms interact simultaneously, with market participants responding and adapting to decarbonization pathways.

电力系统是将当今社会转变为无碳经济的关键。长期电力市场机制，包括拍卖、支持计划和其他政策工具，对形成发电组合至关重要。鉴于需要更先进的工具来支持政策制定者和其他利益相关者设计、测试和评估长期市场，这项工作提出了一个多智能体强化学习模型，能够捕捉脱碳能源系统的关键特征。利润最大化的发电公司根据系统需求、竞争动态和政策信号，在批发电力市场上做出投资决策。该模型采用独立的最近邻策略优化，选择最近邻策略以适应分散的竞争环境。然而，考虑到在多智能体设置中独立学习的固有挑战，广泛的超参数搜索确保分散训练产生与竞争行为一致的市场结果。该模型应用于意大利电力系统的一个程式化版本，并在不同程度的竞争、市场设计和政策场景下进行了测试。结果强调了市场设计对电力部门脱碳和避免价格波动的关键作用。拟议的框架允许评估长期电力市场，其中多种政策和市场机制同时相互作用，市场参与者对脱碳途径作出反应和适应。

{"title":"Assessing long-term electricity market design for ambitious decarbonization targets using multi-agent reinforcement learning","authors":"Javier Gonzalez-Ruiz , Carlos Rodriguez-Pardo , Iacopo Savelli , Alice Di Bella , Massimo Tavoni","doi":"10.1016/j.egyai.2025.100665","DOIUrl":"10.1016/j.egyai.2025.100665","url":null,"abstract":"<div><div>Electricity systems are key to transforming today’s society into a carbon-free economy. Long-term electricity market mechanisms, including auctions, support schemes, and other policy instruments, are critical in shaping the electricity generation mix. In light of the need for more advanced tools to support policymakers and other stakeholders in designing, testing, and evaluating long-term markets, this work presents a multi-agent reinforcement learning model capable of capturing the key features of decarbonizing energy systems. Profit-maximizing generation companies make investment decisions in the wholesale electricity market, responding to system needs, competitive dynamics, and policy signals. The model employs independent proximal policy optimization, which was selected for suitability to the decentralized and competitive environment. Nevertheless, given the inherent challenges of independent learning in multi-agent settings, an extensive hyperparameter search ensures that decentralized training yields market outcomes consistent with competitive behavior. The model is applied to a stylized version of the Italian electricity system and tested under varying levels of competition, market designs, and policy scenarios. Results highlight the critical role of market design for decarbonizing the electricity sector and avoiding price volatility. The proposed framework allows assessing long-term electricity markets in which multiple policy and market mechanisms interact simultaneously, with market participants responding and adapting to decarbonization pathways.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100665"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926797","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

Towards secure federated learning for energy forecasting under adversarial attacks 面向对抗性攻击下能源预测的安全联邦学习

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100680

Jonas Sievers , Krupali Kumbhani , Thomas Blank , Frank Simon , Andreas Mauthe

Federated learning is increasingly used in energy forecasting, enabling buildings to collaboratively predict load, photovoltaic generation, and prosumption while preserving data privacy. However, this collaborative nature introduces new vulnerabilities, as manipulations by a single participant can propagate across the network. Such attacks can undermine grid balancing, limit flexibility provision, and reduce trust in decentralized energy systems. This work presents a comprehensive study of adversarial threats and defenses in federated energy forecasting. We compare structured manipulations generated with Generative Adversarial Networks against simple random perturbations in two attack scenarios: (i) data poisoning, where corrupted training data degrade global accuracy, and (ii) backdoors, where hidden triggers distort predictions in targeted time windows. Our experiments show that poisoning can increase global forecasting errors by up to 131 %, while backdoors raise local errors by up to 48 %. In both cases, Generative Adversarial Network-based attacks are consistently more effective than random perturbations, with backdoors proving especially challenging to detect due to their localized effect. To mitigate these threats, we evaluate four defense strategies: weighted aggregation, larger participant clusters, local retraining, and their coordinated integration into a secure framework. Results demonstrate that these defenses substantially reduce the impact of attacks, and in some cases even improve baseline accuracy, thereby enhancing the resilience of federated energy forecasting against adversarial manipulation.

联邦学习越来越多地用于能源预测，使建筑物能够在保护数据隐私的同时协同预测负荷、光伏发电和消耗。然而，这种协作性引入了新的漏洞，因为单个参与者的操作可以通过网络传播。这种攻击会破坏电网平衡，限制灵活性供应，并降低对分散能源系统的信任。这项工作对联邦能源预测中的对抗性威胁和防御进行了全面研究。我们比较了生成式对抗网络在两种攻击场景下对简单随机扰动产生的结构化操作：(i)数据中毒，其中损坏的训练数据会降低全局准确性，以及（ii）后门，其中隐藏的触发器会扭曲目标时间窗口中的预测。我们的实验表明，中毒可以使全局预测误差增加高达131%，而后门可以使局部误差增加高达48%。在这两种情况下，基于生成对抗网络的攻击始终比随机扰动更有效，由于其局部效应，后门被证明特别具有挑战性。为了减轻这些威胁，我们评估了四种防御策略：加权聚合、更大的参与者集群、本地再培训以及它们协调集成到一个安全框架中。结果表明，这些防御大大减少了攻击的影响，在某些情况下甚至提高了基线准确性，从而增强了联邦能源预测对对抗性操纵的弹性。

{"title":"Towards secure federated learning for energy forecasting under adversarial attacks","authors":"Jonas Sievers , Krupali Kumbhani , Thomas Blank , Frank Simon , Andreas Mauthe","doi":"10.1016/j.egyai.2026.100680","DOIUrl":"10.1016/j.egyai.2026.100680","url":null,"abstract":"<div><div>Federated learning is increasingly used in energy forecasting, enabling buildings to collaboratively predict load, photovoltaic generation, and prosumption while preserving data privacy. However, this collaborative nature introduces new vulnerabilities, as manipulations by a single participant can propagate across the network. Such attacks can undermine grid balancing, limit flexibility provision, and reduce trust in decentralized energy systems. This work presents a comprehensive study of adversarial threats and defenses in federated energy forecasting. We compare structured manipulations generated with Generative Adversarial Networks against simple random perturbations in two attack scenarios: (i) data poisoning, where corrupted training data degrade global accuracy, and (ii) backdoors, where hidden triggers distort predictions in targeted time windows. Our experiments show that poisoning can increase global forecasting errors by up to 131%, while backdoors raise local errors by up to 48%. In both cases, Generative Adversarial Network-based attacks are consistently more effective than random perturbations, with backdoors proving especially challenging to detect due to their localized effect. To mitigate these threats, we evaluate four defense strategies: weighted aggregation, larger participant clusters, local retraining, and their coordinated integration into a secure framework. Results demonstrate that these defenses substantially reduce the impact of attacks, and in some cases even improve baseline accuracy, thereby enhancing the resilience of federated energy forecasting against adversarial manipulation.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100680"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022833","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

A graph neural network enhanced decision transformer for efficient optimization in dynamic smart charging environments 一种基于图神经网络的决策变压器，用于动态智能充电环境的高效优化

IF 9.6 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Energy and AI

Pub Date : 2026-01-01 DOI: 10.1016/j.egyai.2026.100679

Stavros Orfanoudakis, Nanda Kishor Panda, Peter Palensky, Pedro P. Vergara

Electric-vehicle smart charging requires quick decision-making under uncertainty while enforcing strict electricity grid and user requirements. Mathematical optimization becomes too slow at scale, while online reinforcement learning struggles with sparse rewards and safety. This paper proposes GNN-DT, a topology-aware Decision Transformer that combines graph neural network embeddings with sequence modeling to learn charging policies from offline trajectories. The method operates over variable numbers of vehicles and chargers without retraining. Evaluated on realistic smart charging scenarios, GNN-DT achieves near-optimal performance, reaching rewards within 5 percent of an oracle solver while using up to 10

\times

fewer training trajectories than baseline methods. It consistently outperforms online and offline reinforcement learning approaches and generalizes to unseen fleet sizes and network topologies. Inference runs in milliseconds, making the approach suitable for real-time deployment in large-scale charging systems.

电动汽车智能充电要求在不确定条件下快速决策，同时严格执行电网和用户要求。数学优化在规模上变得太慢，而在线强化学习在稀疏的奖励和安全性上挣扎。本文提出了一种拓扑感知决策转换器GNN-DT，它将图神经网络嵌入与序列建模相结合，从离线轨迹中学习充电策略。该方法可以在不需要再培训的情况下在可变数量的车辆和充电器上运行。通过对现实智能充电场景的评估，GNN-DT实现了近乎最佳的性能，在5%的oracle求解器内获得奖励，同时使用的训练轨迹比基线方法少10倍。它始终优于在线和离线强化学习方法，并推广到看不见的车队规模和网络拓扑。推理以毫秒为单位运行，使该方法适合大规模收费系统的实时部署。

{"title":"A graph neural network enhanced decision transformer for efficient optimization in dynamic smart charging environments","authors":"Stavros Orfanoudakis, Nanda Kishor Panda, Peter Palensky, Pedro P. Vergara","doi":"10.1016/j.egyai.2026.100679","DOIUrl":"10.1016/j.egyai.2026.100679","url":null,"abstract":"<div><div>Electric-vehicle smart charging requires quick decision-making under uncertainty while enforcing strict electricity grid and user requirements. Mathematical optimization becomes too slow at scale, while online reinforcement learning struggles with sparse rewards and safety. This paper proposes GNN-DT, a topology-aware Decision Transformer that combines graph neural network embeddings with sequence modeling to learn charging policies from offline trajectories. The method operates over variable numbers of vehicles and chargers without retraining. Evaluated on realistic smart charging scenarios, GNN-DT achieves near-optimal performance, reaching rewards within 5 percent of an oracle solver while using up to 10<span><math><mo>×</mo></math></span> fewer training trajectories than baseline methods. It consistently outperforms online and offline reinforcement learning approaches and generalizes to unseen fleet sizes and network topologies. Inference runs in milliseconds, making the approach suitable for real-time deployment in large-scale charging systems.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"23 ","pages":"Article 100679"},"PeriodicalIF":9.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022834","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