Model-based calibration for battery characterization in HEV applications

Abstract

In Hybrid Electric Vehicle (HEV) applications, unlike electric vehicles, operation with the battery system requires control in a relatively limited range of state-of-charge (SoC), where best efficiency, gradual aging, and no self-damaging operations are expected. In this context, one of the main, critical technical challenges is the estimation of the SoC under vehicle operations, which typically do not involve full charging or discharging. This task is particularly arduous to accomplish in real-time, due to the complex and nonlinear behavior of the battery, as well as the inevitable presence of on-board measurement errors. In this work, we describe a model-based calibration process for capturing the important characteristics of modern batteries used in typical HEV applications. This process consists of reproducible procedural steps, including prespecified data collection, while ultimately admitting a calibration. The resulting models are useful in HEV system control design for algorithms centered on maintaining the battery SoC, in algorithms for prognostics and diagnostics, and in prediction and estimation tasks.