Accurate SOC Prediction and Monitoring of Each Cell in a Battery Pack Considering Various Influencing Factors
Linhui Zhao, Pengliang Qin
Abstract
Accurate prediction and monitoring of the state of charge (SOC) of each cell in a battery pack are of great significance for safe driving of electric vehicles. Subject to the factors of differences between cells, temperature, and aging, the accuracy of existing SOC prediction methods may be affected in applications. This article employs a data-driven method with a proposed novel transfer learning framework considering conditional probability distribution adaptation to solve the impact of the aforementioned influencing factors on SOC prediction and obtains a favorable SOC prediction result for each cell. As experiments of actual battery pack provided by China First Automobile Work based demonstrated, the proposed method can accurately predict SOC of each cell under different influencing factors by using only one cell modeling data. Moreover, the proposed method is robust against model parameter uncertainties, sensor noise, etc.