Energy and AI最新文献_第5页

Harnessing distributed GPU computing for generalizable graph convolutional networks in power grid reliability assessments

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2025.100471

Somayajulu L.N. Dhulipala , Nicholas Casaprima , Audrey Olivier , Bjorn C. Vaagensmith , Timothy R. McJunkin , Ryan C. Hruska

{"title":"Harnessing distributed GPU computing for generalizable graph convolutional networks in power grid reliability assessments","authors":"Somayajulu L.N. Dhulipala , Nicholas Casaprima , Audrey Olivier , Bjorn C. Vaagensmith , Timothy R. McJunkin , Ryan C. Hruska","doi":"10.1016/j.egyai.2025.100471","DOIUrl":"10.1016/j.egyai.2025.100471","url":null,"abstract":"<div><div>Although machine learning (ML) has emerged as a powerful tool for rapidly assessing grid contingencies, prior studies have largely considered a static grid topology in their analyses. This limits their application, since they need to be re-trained for every new topology. This paper explores the development of generalizable graph convolutional network (GCN) models by pre-training them across a wide range of grid topologies and contingency types. We found that a GCN model with auto-regressive moving average (ARMA) layers with a line graph representation of the grid offered the best predictive performance in predicting voltage magnitudes (VM) and voltage angles (VA). We introduced the concept of phantom nodes to consider disparate grid topologies with a varying number of nodes and lines. For pre-training the GCN ARMA model across a variety of topologies, distributed graphics processing unit (GPU) computing afforded us significant training scalability. The predictive performance of this model on grid topologies that were part of the training data is substantially better than the direct current (DC) approximation. Although direct application of the pre-trained model to topologies that are not part of the grid is not particularly satisfactory, fine-tuning with small amounts of data from a specific topology of interest significantly improves predictive performance. In the context of foundational models in ML, this paper highlights the feasibility of training large-scale GNN models to assess the reliability of power grids by considering a wide variety of grid topologies and contingency types.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100471"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Robust survival model for the prediction of Li-ion battery lifetime reliability and risk functions

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100465

Rasheed Ibraheem , Timothy I. Cannings , Torben Sell , Gonçalo dos Reis

{"title":"Robust survival model for the prediction of Li-ion battery lifetime reliability and risk functions","authors":"Rasheed Ibraheem , Timothy I. Cannings , Torben Sell , Gonçalo dos Reis","doi":"10.1016/j.egyai.2024.100465","DOIUrl":"10.1016/j.egyai.2024.100465","url":null,"abstract":"<div><div>Single-value prediction such as the End of Life and Remaining Useful Life is a common method of estimating the lifetime of Li-ion batteries. Information from such prediction is limited when the entire degradation pattern is needed for practical applications such as dynamic adjustment of battery warranty, improved maintenance scheduling, and battery stock management. In this research, a predictive, semi-parametric survival model called the Cox Proportional Hazards is proposed for the prediction of cell degradation in the form of survival probability (battery reliability) and cumulative hazard (battery risk) functions. Once this model is trained, the two functions can be obtained directly for a new cell without having to predict several cogent points. The model is trained on the first 50 cycles of only the voltage profile from either the charge or discharge data regime, implying that our methodology is data region agnostic. The signature method with both desirable mathematical and machine learning properties was adopted as a feature extraction technique.</div><div>The developed models are tested rigorously using application-driven strategies involving model robustness to the number of cycles of data required for model training and prediction, different fractions of training samples, and systematic data sparsity. The codes for modeling and testing are publicly available.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100465"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Novel intelligent adaptive sliding mode control for marine fuel cell system via hybrid algorithm

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100464

Shiyi Fang, Daifen Chen, Xinyu Fan

引用次数: 0

Topologically consistent regression modeling exemplified for laminar burning velocity of ammonia-hydrogen flames

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100456

Hui Du , Tianyu Wang , Haogang Wei , Guy Y. Cornejo Maceda , Bernd R. Noack , Lei Zhou

{"title":"Topologically consistent regression modeling exemplified for laminar burning velocity of ammonia-hydrogen flames","authors":"Hui Du , Tianyu Wang , Haogang Wei , Guy Y. Cornejo Maceda , Bernd R. Noack , Lei Zhou","doi":"10.1016/j.egyai.2024.100456","DOIUrl":"10.1016/j.egyai.2024.100456","url":null,"abstract":"<div><div>Data-driven regression models are generally calibrated by minimizing a representation error. However, optimizing the model accuracy may create nonphysical wiggles. In this study, we propose topological consistency as a new metric to mitigate these wiggles. The key enabler is Persistent Data Topology (PDT) which extracts a topological skeleton from discrete scalar field data. PDT identifies the extrema of the model based on a neighborhood analysis. The topological error is defined as the mismatch of extrema between the data and the model. The methodology is exemplified for the modeling of the Laminar Burning Velocity (<span><math><mrow><mi>L</mi><mi>B</mi><mi>V</mi></mrow></math></span>) of ammonia-hydrogen flames. Four regression models, Multi-layer Perceptron (MLP), eXtreme Gradient Boosting (XGBoost), Random Forest (RF), and Light Gradient Boosting Machine (Light GBM), are trained using the data generated by a modified GRI3.0 mechanism. In comparison, MLP builds a model that achieves the highest accuracy and preserves the topological structure of the data. We expect that the proposed topologically consistent regression modeling will enjoy many more applications in model calibration, model selection and optimization algorithms.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100456"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155831","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Probabilistic forecasting of multiple plant day-ahead renewable power generation sequences with data privacy preserving

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100463

Hong Liu , Zijun Zhang

{"title":"Probabilistic forecasting of multiple plant day-ahead renewable power generation sequences with data privacy preserving","authors":"Hong Liu , Zijun Zhang","doi":"10.1016/j.egyai.2024.100463","DOIUrl":"10.1016/j.egyai.2024.100463","url":null,"abstract":"<div><div>This paper studies the renewable power forecasting task with a more advanced formulation, the probabilistic forecasts of day-ahead power generation sequences of multiple renewable power plants without breaching the privacy of data in each plant. To realize such a task, an advanced domain-invariant feature learning embedded federated learning (DIFL) framework is proposed to coordinate the development of a system of deep network-based models serving as multiple clients and one server. In DIFL, each client, which serves each local renewable power plant, maps its raw data input into latent features via a local feature extractor and generates power output sequence probabilistic forecasts via a locally hosted forecasting model. The cloud-hosted server first aggregates the knowledge from models of clients and next dispatches the aggregated model back to each client for facilitating each local feature extractor to identify domain-invariant features via interacting with a server-side discriminator. Therefore, only desensitized data, such as parameters of the models, are allowed to be transmitted among end users for preserving local data privacy of power plants. To verify the advantages of the DIFL, a preliminary exploration of its theoretical property is first conducted. Next, computational studies are performed to benchmark the DIFL against famous baselines based on datasets collected from commercial renewable power plants. Results further confirm that, in terms of the averaged performance, the DIFL consistently realizes improvements against all benchmarks based on both real wind farm and solar power plant datasets.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100463"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155836","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Using crafted features and polar bear optimization algorithm for short-term electric load forecast system

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2025.100470

Mansi Bhatnagar, Gregor Rozinaj, Radoslav Vargic

{"title":"Using crafted features and polar bear optimization algorithm for short-term electric load forecast system","authors":"Mansi Bhatnagar, Gregor Rozinaj, Radoslav Vargic","doi":"10.1016/j.egyai.2025.100470","DOIUrl":"10.1016/j.egyai.2025.100470","url":null,"abstract":"<div><div>Short-term load forecasting (STLF) can be utilized to predict usage fluctuation in a short time period and accurate forecasting can save a big chunk of a country's economic loss. This paper introduces the crafting of various features for hourly electric load forecasting on three different datasets using four different models XGBoost, LightGBM, Bi-LSTM, and Random Forest. The importance of crafted features over basic features was analysed by different evaluation metrics MAE, RMSE, R-squared, and MAPE. Evaluation metrics showed that prediction accuracy increased significantly with crafted features in comparison to basic features for all four models. We also showcased the ability of the Polar Bear Optimisation (PBO) algorithm for hyperparameter tuning of the machine learning models in STLF. Optimized hyperparameters with PBO effectively decreased RMSE, MAE, and MAPE and improved the model prediction, showcasing the capability of the PBO in hyperparameter tuning for STLF. PBO was compared with commonly used optimization algorithms like particle swarm optimization (PSO) and genetic algorithm (GA). GA was the least performing with XGBoost, LightGBM, and Random Forest. PSO and PBO were comparable with XGBoost LightGBM and Random Forest while PBO highly surpassed PSO with the Bi-LSTM model. Hence PBO was proved to be highly effective for hyperparameter tuning for implementation in short-term electric load forecasting.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100470"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A co-estimation framework of state of health and remaining useful life for lithium-ion batteries using the semi-supervised learning algorithm

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100458

Xiaoyu Li , Mohan Lyv , Xiao Gao , Kuo Li , Yanli Zhu

{"title":"A co-estimation framework of state of health and remaining useful life for lithium-ion batteries using the semi-supervised learning algorithm","authors":"Xiaoyu Li , Mohan Lyv , Xiao Gao , Kuo Li , Yanli Zhu","doi":"10.1016/j.egyai.2024.100458","DOIUrl":"10.1016/j.egyai.2024.100458","url":null,"abstract":"<div><div>To ensure the safe operation of lithium-ion batteries, it is crucial to accurately predict their state of health (SOH) and remaining useful life (RUL). Addressing the issue of high costs and time consumption due to the reliance on large amounts of labeled data in existing models, this paper proposes a co-estimation framework that combines semi-supervised learning (SSL) with long short-term memory networks (LSTM), effectively utilizing unlabeled data. By selecting the most strongly correlated battery health features and constructing a degradation model using a hybrid dataset, the need for labeling is reduced. The verification results indicate SOH estimated error is reduced to 4 % and the maximum root mean square error (RMSE) is 1.58 %. When utilizing 75 % SOH as the end-of-life criterion for battery cycle life, the mean absolute error (MAE) of the RUL predictions for the two tested batteries are 2.5281 and 0.0562 cycles, respectively. The results prove the framework enables accurate prediction and has wide practicability and universal applicability.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100458"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Emission prediction and optimization of methanol/diesel dual-fuel engines based on ITransformer-BiGRU and NSGA-III

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100466

Mingzhang Pan , Xinxin Cao , Changcheng Fu , Shengyou Liao , Xiaorong Zhou , Wei Guan

{"title":"Emission prediction and optimization of methanol/diesel dual-fuel engines based on ITransformer-BiGRU and NSGA-III","authors":"Mingzhang Pan , Xinxin Cao , Changcheng Fu , Shengyou Liao , Xiaorong Zhou , Wei Guan","doi":"10.1016/j.egyai.2024.100466","DOIUrl":"10.1016/j.egyai.2024.100466","url":null,"abstract":"<div><div>To reduce engine pollutant emissions, an emission modeling and optimization scheme based on a hybrid artificial intelligence scheme is proposed in this study to reduce pollutant emissions of methanol/diesel dual-fuel engines under low load. Firstly, a data cleaning method based on isolated forest and correlation analysis is designed to improve the stability of the system. Secondly, a hybrid emission prediction model based on improved Transformer (ITransformer) and Bidirectional Gated Recurrent Unit (BiGRU) is built to obtain an accurate mathematical model between control parameters and emissions. Finally, based on the obtained mathematical model, the 3rd Non-dominated Sorting Genetic Algorithm (NGSA-III) is used to adjust and optimize the control parameters. Using engine bench test data to evaluate the proposed hybrid emission prediction model, the R<sup>2</sup> of CO, HC, and NO<sub>x</sub> prediction is 0.9969, 0.9973, and 0.9982, respectively, which is higher than the accuracy of the seven existing modeling methods. Compared with the unoptimized MESR46, the CO, HC, and NO<sub>x</sub> emissions of the optimized scheme are reduced by at least 45.17 %, 15.30 %, and 17.32 % respectively, which can significantly reduce the CO, HC, and NO<sub>x</sub> emissions, and comparison and analysis with the most advanced optimization technologies show a competitive optimization effect.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100466"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Identification of cell chemistries in lithium-ion batteries: Improving the assessment for recycling and second-life

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100468

Christopher Wett , Jörg Lampe , Dominik Görick , Thomas Seeger , Bugra Turan

{"title":"Identification of cell chemistries in lithium-ion batteries: Improving the assessment for recycling and second-life","authors":"Christopher Wett , Jörg Lampe , Dominik Görick , Thomas Seeger , Bugra Turan","doi":"10.1016/j.egyai.2024.100468","DOIUrl":"10.1016/j.egyai.2024.100468","url":null,"abstract":"<div><div>Recycling and second life of lithium-ion batteries are vital for lowering the growing resource demand of sectors like mobility or home energy storage. However, an often-overlooked issue is the sometimes-unknown cell chemistry of batteries entering the end-of-life. In this work, a machine learning based approach for the identification of lithium-ion battery cathode chemistries is presented. First, an initial measurement boundary determination is introduced. Using the Python Battery Mathematical Modelling (PyBaMM) framework, synthetical partial open circuit voltage (OCV) charge and discharge curves are generated with an electrochemical single particle model for three different cathode chemistries and the initial state of charge and state of health values as well as the initial capacities are varied. The dV/dQ characteristics are chosen as features and four machine learning algorithms are trained on different lengths of OCV curves. The trade-off between achievable accuracy and the number of OCV steps showed that an increasing accuracy correlates with a higher step number. While extremely small charge and discharge capacities per step did not yield sufficient testing accuracies, capacities starting from 0.2 Ah per step up to 0.6 Ah per step showed increasingly good results with an accuracy of up to 89.3 % for 0.5 Ah and 15 OCV steps. Additionally, the approach was validated by classifying experimental data. The results especially demonstrate the effectiveness of the approach to distinguish between lithium iron phosphate (LFP) and lithium nickel manganese cobalt (NMC) cells.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100468"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155830","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A power extraction approach with load state modification for energy disaggregation

IF 9.6

Energy and AI Pub Date : 2025-01-01 DOI: 10.1016/j.egyai.2024.100461

Yusen Zhang , Feng Gao , Kangjia Zhou , Shuquan Wang , Hanzhi Wang

{"title":"A power extraction approach with load state modification for energy disaggregation","authors":"Yusen Zhang , Feng Gao , Kangjia Zhou , Shuquan Wang , Hanzhi Wang","doi":"10.1016/j.egyai.2024.100461","DOIUrl":"10.1016/j.egyai.2024.100461","url":null,"abstract":"<div><div>Energy disaggregation is a technology that disassembles the energy consumption from the entire house into load-level contributions. One of the foundational tasks for this technology is to accurately ascertain the truth electrical energy consumption of the target load. However, current energy disaggregation methods find it difficult to accurately predict the actual operating power of appliances when there are significant differences in the data distribution of appliances across various scenarios due to the diversity in manufacturers, usage times, and operating conditions. In this study, we propose a power extraction approach with load state modification to capture accurate load operating power with minimal influence from usage scenarios. To be specific, the on/off state sequence of appliances is first predicted leveraging existing energy disaggregation methods, and two state modification methods based on non-operating time and operating time of appliances are respectively proposed to modify the erroneous states in sequence. Subsequently, the power extraction approach calculates the operational power of target appliance based on the amplitude of fluctuations within the aggregated energy consumption caused by its state changes. Furthermore, a removing signal spikes method is proposed to improve the accuracy of the extracted power value. We conducted extensive experiments on a public dataset, demonstrating that the proposed method can significantly improve the accuracy of state-of-the-art solution. The average of mean absolute error across commonly used appliances during on state were reduced by 44.75 % and 32.07 % respectively in the UK-DALE and REFIT datasets.</div></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"19 ","pages":"Article 100461"},"PeriodicalIF":9.6,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143155835","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0