Vital health indicator based state of health estimation of lithium-ion battery by adaptive neuro-fuzzy inference system

Abstract

ABSTRACT The battery management system (BMS) in electric vehicles (EVs) relies on State of Health (SOH) assessment to monitor battery health and ensure performance. BMS can detect and prevent cell balance, thermal runaway, and extend its lifespan. Methodical charging and discharging under different circumstances reduce battery degradation. Grid Partitioning (GP), Subtractive Clustering (SC), and Fuzzy C-Means (FCM) from the Adaptive Neuro-Fuzzy Inference System (ANFIS) have been employed to predict battery SOH in this work. This study examines charge-discharge cycles of 8 lithium-ion pouch cells from the Oxford Battery Degradation Dataset. Charging cycle data extracts the five vital health indicators (VHIs), whereas discharging cycle data calculates cell capacity. For analyzing the effects of VHIs on SOH, a correlation matrix is formed. Three instances are defined depending on the individual VHIs impacts. Instance 1 utilizes VHI 1, VHI 2, and VHI 5, instance 2 utilizes VHI 3 and VHI 4, and instance 3 uses all VHIs as inputs. Due to enormous data points, only even-numbered cell data is utilized for training the ANFIS model, while odd-numbered cell data is used for validation and testing. In the best-case scenario, instance 1 gives the lowest root mean square error (RMSE) from all the methods.