Sensitivity analysis of Kalman Filter based capacity estimation for electric vehicles

In this paper, a sensitivity analysis of Kalman Filter based capacity estimation for electric vehicle batteries is performed. In order to represent different user driving behaviours a random drive cycle data generator is introduced and used in the analysis. An asymmetric equivalent circuit is used for both data generation and estimation. Different aspects and sources of measurement and modeling error are introduced to investigate their impact on accuracy. It is demonstrated the nonlinear nature of the open circuit voltage and modeling errors of its curvature are among the greatest sources of estimation error. Additional analysis to study effects of error resulting from measurement error, discretization, and smoothing are also presented.