| Title |
State-of-Health Estimation of Lithium-Ion Battery based on Convolutional Neural Network considering Health Indicator Extraction |
| Authors |
문태석(Tae-Suk Mun) ; 한동호(Dong-Ho Han) ; 권상욱(Sang-Uk Kwon) ; 백종복(Jong-Bok Baek) ; 김종훈(Jong-Hoon Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2021.70.10.1467 |
| Keywords |
Health indicator; Lithium-ion battery; Convolutional neural network; Urban dynamometer driving schedule; State-of-health estimation |
| Abstract |
Due to the output characteristics of lithium-ion batteries used in electric vehicles (EVs) vary according to battery aging, accurate prediction of the state-of-health (SOH) reflecting the aging state is important. However, it is difficult to consider factors affecting battery characteristics, such as frequent charging and discharging, operating temperature, and state of charge, so there is a limitation in predicting battery SOH. Even in a model that considers the aging state of the battery under these various conditions, there is also a problem in that the complexity of the calculation process and parameters becomes serious. Therefore, in this paper, in order to research on the estimation of the aging state during operation of the INR18650-25R battery, a health indicator (HI) that can reflect the internal state of the battery according to aging is extracted. This research produce a learning image through the extracted HI and build a model study that enables algorithm learning through the image. The experimental profile used for model training and validation was an Urban Dynamometer Driving Schedule (UDDS), and a Convolutional Neural Network (CNN) with strength in image learning was applied for the estimation algorithm. |