Thomas Kröger , Annalena Belnarsch , Philip Bilfinger , Wolfram Ratzke , Markus Lienkamp
{"title":"Collaborative training of deep neural networks for the lithium-ion battery aging prediction with federated learning","authors":"Thomas Kröger , Annalena Belnarsch , Philip Bilfinger , Wolfram Ratzke , Markus Lienkamp","doi":"10.1016/j.etran.2023.100294","DOIUrl":null,"url":null,"abstract":"<div><p><span><span>Accurate and reliable prediction of the future capacity degradation of lithium-ion batteries is crucial for their application in electric vehicles. Recent publications have highlighted the effectiveness of deep learning, in particular, in generating precise forecasts regarding the aging patterns. However, large quantities of training data covering various aging behaviors are required to train such models effectively. Collecting such a large database centrally is not feasible due to privacy and data communication restrictions of data owners, such as testing facilities or fleet operators. Federated learning provides a solution to this open issue. A framework, which incorporates federated learning into the training of a data-based </span>battery aging model, is presented in this paper. The benefit of federated learning is that even data owners with sensible information can participate in a collaborative model training, since the model training is only conducted locally and all the data remains local and does not have to be disclosed. Thus, more data owners are likely to participate in this collaborative training. This will improve the prediction performance due to the enlarged dataset that can be utilized for the model training. This work shows that the prediction accuracy of the model trained with federated learning is only slightly worse than the prediction results obtained by the ideal case in which all aging data is stored in a central database. A sensitivity analysis is presented to prove the robustness of federated learning even if the datasets between participating data owners are highly imbalanced or exhibit different aging behaviors. Within exemplary scenarios, it is shown that individual data holders can reduce their prediction errors from </span><span><math><mrow><mi>M</mi><mi>A</mi><mi>P</mi><msub><mrow><mi>E</mi></mrow><mrow><mi>m</mi><mi>e</mi><mi>a</mi><mi>n</mi></mrow></msub><mo>=</mo><mn>7</mn><mo>.</mo><mn>07</mn><mtext>%</mtext></mrow></math></span> to <span><math><mrow><mi>M</mi><mi>A</mi><mi>P</mi><msub><mrow><mi>E</mi></mrow><mrow><mi>m</mi><mi>e</mi><mi>a</mi><mi>n</mi></mrow></msub><mo>=</mo><mn>0</mn><mo>.</mo><mn>91</mn><mtext>%</mtext></mrow></math></span> by participating in the proposed federated learning-based framework.</p></div>","PeriodicalId":36355,"journal":{"name":"Etransportation","volume":"18 ","pages":"Article 100294"},"PeriodicalIF":15.0000,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Etransportation","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2590116823000693","RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0
Abstract
Accurate and reliable prediction of the future capacity degradation of lithium-ion batteries is crucial for their application in electric vehicles. Recent publications have highlighted the effectiveness of deep learning, in particular, in generating precise forecasts regarding the aging patterns. However, large quantities of training data covering various aging behaviors are required to train such models effectively. Collecting such a large database centrally is not feasible due to privacy and data communication restrictions of data owners, such as testing facilities or fleet operators. Federated learning provides a solution to this open issue. A framework, which incorporates federated learning into the training of a data-based battery aging model, is presented in this paper. The benefit of federated learning is that even data owners with sensible information can participate in a collaborative model training, since the model training is only conducted locally and all the data remains local and does not have to be disclosed. Thus, more data owners are likely to participate in this collaborative training. This will improve the prediction performance due to the enlarged dataset that can be utilized for the model training. This work shows that the prediction accuracy of the model trained with federated learning is only slightly worse than the prediction results obtained by the ideal case in which all aging data is stored in a central database. A sensitivity analysis is presented to prove the robustness of federated learning even if the datasets between participating data owners are highly imbalanced or exhibit different aging behaviors. Within exemplary scenarios, it is shown that individual data holders can reduce their prediction errors from to by participating in the proposed federated learning-based framework.
期刊介绍:
eTransportation is a scholarly journal that aims to advance knowledge in the field of electric transportation. It focuses on all modes of transportation that utilize electricity as their primary source of energy, including electric vehicles, trains, ships, and aircraft. The journal covers all stages of research, development, and testing of new technologies, systems, and devices related to electrical transportation.
The journal welcomes the use of simulation and analysis tools at the system, transport, or device level. Its primary emphasis is on the study of the electrical and electronic aspects of transportation systems. However, it also considers research on mechanical parts or subsystems of vehicles if there is a clear interaction with electrical or electronic equipment.
Please note that this journal excludes other aspects such as sociological, political, regulatory, or environmental factors from its scope.