EMD-KPCA based Short Circuit Fault Diagnosis Method for Battery Pack

Abstract

In this article, a new sensor topology and signal processing method for battery pack faults are proposed. First, measure the voltage of each cell, calculate the recursive correlation coefficient (CC) of adjacent cell voltages to detect whether there is a fault in the battery pack, and form a new data matrix. Then, various feature indicators are extracted from the new matrix by classical modal decomposition, and the most representative features are extracted by Kernel Principal Component Analysis (KPCA). Finally, the support vector machine is used to build a fault classification model to identify fault features. Physically trigger internal short-circuit faults on battery packs connected in series to obtain realistic datasets. A comparison of different algorithms under the same conditions shows that the proposed fault diagnosis method can accurately and reliably predict internal short-circuit faults, with a fault detection success rate of 80%.