Robust Data-Driven Error Compensation for a Battery Model

Abstract Models of traction batteries are an essential tool throughout the development automotive drivetrains. Surprisingly, today’s massively collected battery data is not yet used for more accurate and reliable simulations. Primarily, non-uniform excitation during regular operations prevent a consequent utilization such measurements. Hence, there need methods which enable robust models based on large datasets. For that reason, data-driven error model introduced enhancing existing physically motivated model. A neural network compensates dynamic further limited description underlying data. This paper tries to verify effectiveness robustness general setup additionally evaluates one-class support vector machine as proposed training distribution. Based five datasets it shown, gradually limiting compensation outside boundary leads similar improvement increased overall robustness.