Prognostics for State of Health Estimation of Battery System Under Uncertainty Based on Adaptive Learning Technique

This paper addresses the State of health (SOH) estimation for system in the prognostics and health management (PHM) of the electronics systems. Due to complicated operation conditions, it is necessary to derive and implement the prognostics with uncertain situations. In this paper, a Bayesian filtering approach based on the adaptive learning for the Gaussian process regression (GPR) model is presented for the SOH estimation of system under uncertain conditions. Instead of assuming the certain state space model for the degradation trend directly, the distribution of the degradation process is investigated to learn from the inputs based on available measurements. In order to capture the time-varying degradation behavior, the proposed method represents the statistical property of the degradation process through distribution learning with the GPR model. By exploiting the distribution information of degradation process, the particle filter can be implemented to predict the SOH of system. Experiments and comparison analysis are provided to demonstrate the efficiency of the proposed approach.