Online Health Estimate of Hybrid Energy Storage System Based on Fuzzy Brain Emotional Learning Neural Networks

This paper aims to propose a more efficient estimator and applies it to health estimation for battery and supercapacitor in the hybrid energy storage system (HESS). A novel method for online health estimator based on fuzzy brain emotional learning neural network (FBELNN) is proposed. It’s different from conventional fuzzy brain emotional learning neural network that the fuzzy inference system and the novel reward signal are applied in this paper. The proposed method uses wavelet packet decomposition (WPD) and principal component analysis (PCA) to extract features from impulse respond of load surges. The parameter adaptation laws of the FBELNN are derived and wavelet packet function selection method based on the frequency band energy entropy is presented. The method of WPD-PCA can reduce the workload of feature extraction. Through the neural network estimate the capacity of battery and supercapacitor in real-time, one can better ensure the safety of HESS. The training samples and test samples are collected from the response of voltage signals in HESS simulation platform. Compared to other conventional methods, it’s shown that the anti-noise performance and the accuracy are improved by the proposed method.