Fault diagnosis for electric vehicle lithium batteries using a multi-classification support vector machine

To overcome the complexity of fault diagnosis in electric vehicle batteries and the challenges in obtaining fault state data, we propose a fault diagnosis method based on a multi-classification support vector machine (MC-SVM). This approach decreases the dependence on data volume while increasing the diagnosis accuracy and speed. Kernel function parameters were optimized. The experimental results showed that our method improved training speeds and accuracy compared to Product-based Neural Network (PNN). The proposed approach provides a promising result in diagnosing electric vehicle battery fault with small sample training sets. It could increase the safety and efficiency of electric vehicle battery systems.