Hybrid Deep Neural Network for Electric Vehicle State of Charge Estimation

Basit Öğe Kaydı
dc.contributor.authorKadem, Onur
dc.contributor.authorCandan, Hasibe
dc.contributor.authorKim, Jongrae H.
dc.date.accessioned2026-02-08T15:11:01Z
dc.date.available2026-02-08T15:11:01Z
dc.date.issued2024
dc.departmentBursa Teknik Üniversitesi
dc.description3rd IEEE International Conference on Electrical Power and Energy Systems, ICEPES 2024 -- 2024-06-21 through 2024-06-22 -- Bhopal -- 202357
dc.description.abstractIn electric vehicles (EVs), a battery management system plays a critical role in ensuring the reliable and safe operation of batteries. One of its main tasks is to display the battery state of charge (SoC), reflecting the battery's current available charge level. However, the accuracy of SoC estimation is a formidable challenge due to the intricate nature of battery modelling. To overcome this challenge, data-driven methods have recently emerged as the dominant approach for SoC estimation. Considering the SoC estimation problem as a time series problem, we propose a hybrid deep neural network (DNN) that eliminates the need for feature engineering or adaptive filtering. The proposed DNN incorporates a convolutional layer, a long short-term memory layer, and a dense layer. The DNN was trained using data collected from benchmark EV driving cycles (DST, BJDST, and FUDS drive cycles) within a temperature range of 0 °C to 50 °C. The performance evaluation of the trained DNN has been carried out using another standard EV driving cycle (US06 drive cycle) at various operating temperatures. The results demonstrate that the trained DNN effectively captures the dynamic behaviour of the battery under various operational conditions, yielding a maximum percentage SoC estimation error of approximately 3%. Furthermore, the results indicate that the proposed DNN technique is capable of generalising the battery's dynamic response to unseen data. Overall, our findings show that the proposed technique is promising for EV applications in which battery operating conditions exhibit significant variability. © 2024 IEEE.
dc.description.sponsorshipIEEE; IEEE MP Section; IEEE PELS/IES MP Section; IEEE PES MP Section; IETE MP Section
dc.identifier.doi10.1109/ICEPES60647.2024.10653585
dc.identifier.isbn9798350390728
dc.identifier.scopus2-s2.0-85204286076
dc.identifier.scopusqualityN/A
dc.identifier.urihttps://doi.org/10.1109/ICEPES60647.2024.10653585
dc.identifier.urihttps://hdl.handle.net/20.500.12885/5164
dc.indekslendigikaynakScopus
dc.language.isoen
dc.publisherInstitute of Electrical and Electronics Engineers Inc.
dc.relation.publicationcategoryKonferans Öğesi - Uluslararası - Kurum Öğretim Elemanı
dc.rightsinfo:eu-repo/semantics/closedAccess
dc.snmzScopus_KA_20260207
dc.subjectBattery management system
dc.subjectConvolutional neural network
dc.subjectElectrical battery
dc.subjectLong short term memory
dc.subjectState of charge estimation
dc.titleHybrid Deep Neural Network for Electric Vehicle State of Charge Estimation
dc.typeConference Object

Dosyalar

Koleksiyon

Scopus İndeksli Yayınlar Koleksiyonu