Multifault Diagnosis Using the PSO-ET With a Smaller Number of Sensors for Electric Vehicles

The timely diagnosis of the fault type and location information of lithium batteries is necessary during the application of electric vehicles (EVs). Therefore, in this study, a topology was devised to reduce the application of sensors in the structure by producing uneven changes in the signals within the topology when internal short circuit (ISC) and cell open circuit (COC) faults occurred. Further, this characteristic information, which was obtained after the occurrence of the fault, was used to accurately capture the fault information using particle swarm optimization–extra trees (PSO–ET). First, the features that could be used for fault information diagnosis were extracted. The extracted fault information was diagnosed using PSO-ET to obtain information on COC or ISC faults and locations. The proposed method was validated using datasets under different types of working conditions. The proposed method in this article is capable of reducing the number of sensors used in a 16-cell battery pack by 13, while ensuring a prompt and effective response time for detecting and locating faults.