State of Health Estimation of Lithium-ion Battery Using a CS-SVR Model Based on Evidence Reasoning Rule

Abstract

The state of health (SOH) estimation accuracy of lithium-ion battery affects the safety and service life of batteries. Aimed at the problem in SOH estimation of lithium-ion battery, a cuckoo search support vector regression (CS-SVR) model based on the evidence reasoning (ER) rule was proposed for SOH estimation. The lithium-ion battery data from NASA Ames Center was used to perform the SOH estimation test. In this method, the average voltage and average temperature of battery discharge cycles were taken as model input, and a fusion belief degree matrix of input data was obtained by the ER rule. The SOH estimation result of the battery was obtained by inputting a fusion belief degree matrix into the SVR model optimized by the CS algorithm. The results show that the CS-SVR algorithm based on the ER rule has a better estimation performance than the five existing models.

Keywords

• lithium-ion battery
• evidence reasoning (er) rule
• cuckoo search (cs)
• state of health (soh) estimation
• support vector regression (svr)