Abnormal sensing feature detection of DC high voltage power battery for new energy vehicles

Abstract

As a kind of clean energy transportation, new energy vehicles are widely respected. This topic focuses on the detection of abnormalities in power batteries in new energy vehicles. After combing the common faults of the battery management system, using the basic structure of RBF neural network and the advantages of the reduced clustering algorithm, for a single power battery, the power battery power abnormality detection scheme based on the improvement of reduced clustering algorithm is proposed, and the power battery abnormality detection process is designed. Taking the sensing feature data of the battery management system of a new energy vehicle as an experimental sample, through the battery state estimation experiment and the example application of the model, it is found that the RMSE (0.0018) and MAPE (0.0206) of the model training are lower than that of the comparison model, and the average error rate of the abnormal battery identification is 0.833%. The model’s abnormality detection results in both instances are consistent with the actual maintenance results. The analysis indicates that the RBF neural network model with reduced clustering algorithm has superior accuracy and feasibility for detecting abnormal battery power.

Keywords

• reduced clustering algorithm
• rbf neural network
• anomaly detection
• state estimation
• power battery
• 00a79