Research On Signal Characteristics And Diagnosis Methods Of Series Arc Fault In Electric Vehicle Battery Packs

Electric vehicles have the advantages of energy saving and environmental protection,low noise,simple structure,low price,and fast acceleration.The battery pack is a key component of electric vehicles,and its working reliability can directly affect the safety of electric vehicles.Electric vehicle fire accidents are mostly caused by battery pack failures,among which arc faults are the main cause of battery fires,but there is currently a lack of effective detection methods.This article investigated the failure of the current interrupt device(CID)of the cell and the loose contact under vibration.An electric vehicle battery pack DC arc fault generator was developed to simulate the internal DC arc fault of the battery pack,and an electric vehicle battery pack arc fault experiment system was built based on the load characteristics of the electric vehicle.The arc fault experiments was carried out under different arcing reasons,different vibration frequencies,and different motor operating states,and a large amount of experimental data was collected.By using deep learning,three DC arc fault detection methods for electric vehicle battery packs was proposed:The RNN diagnosis model based on time domain signal;d-vector diagnosis model based on RNN model;and the CNN diagnosis model based on spectrogram are proposed.It not only provides a theoretical basis for improving the battery management system of electric vehicles,but also provides technical means for timely detection of DC arc faults in the battery packs and elimination of hidden troubles,and improves the safety and reliability of electric vehicles.