A novel image recognition based fault diagnostics of customized EV battery pack using optimized GoogLeNet

Abstract Diagnosing faults in Battery Management Systems is critical in ensuring rechargeable batteries’ safe and efficient operation. Various fault detection techniques have been developed to address this requirement. This paper focuses on a novel approach to fault diagnosis by utilizing a personalized battery pack and employing image recognition. This study collects three distinct types of fault data from the battery pack and transforms it into image format. These images are inputs to a sophisticated Optimized GoogLeNet model designed explicitly for fault detection in the Battery Management System. The model’s capability to accurately detect and classify different types of faults in the battery pack is thoroughly evaluated. The results generated by the developed model are compared with those produced by the AlexNet and GoogLeNet models. It is demonstrated that the Optimized GoogLeNet model exhibits superior performance in fault detection by obtaining an accuracy of 99.5%, marking a significant improvement over existing techniques. The findings establish that leveraging image-based techniques with advanced Deep learning algorithms holds promising potential for enhancing the accuracy and efficacy of fault diagnosis in personalized battery packs.