A SOC estimation method based on Improved Fuzzy Broad Learning System

Abstract

In this paper, FBLS is used as the base model and Grasshopper optimization Algorithm algorithm(GOA) is used to find the optimal initial weight of Fuzzy Broad Learning System(FBLS). On the basis of finding the initial center of membership function, K-means algorithm is improved by using local density and local probability theory to find the optimal initial clustering center, so that the algorithm can be more accurate. According to the disadvantage that Grasshopper optimization Algorithm(GOA) is not easy to converge and jump out of the global optimization, the algorithm is improved by using the sine cosine theory, and the algorithm search is more comprehensive by improving the parameter Rl. The experimental simulation shows that the error of the improved SC-GOA-FBLS is significantly lower than that of GOA-FBLS,PSO-BP and FBLS.