Electric Vehicle Battery Fault Diagnosis Based on Statistical Method. (May 2017)

• Record Type: Journal Article
• Authors: Zhao, Yang; Liu, Peng; Wang, Zhenpo; Hong, Jichao
• Is Part Of: Energy procedia. Volume 105(2017)
• Page Start: 2366
• Page End: 2371
• Publication Date: 2017-05
• Subjects: Electric vehicle -- Power battery system -- Fault diagnosis -- Big data statistics -- Neural network; Power resources -- Congresses; Power resources -- Periodicals; Power resources; Conference proceedings; Periodicals; 333.7905
• DOI: 10.1016/j.egypro.2017.03.679
• Languages: English
• ISSNs: 1876-6102

Abstract

Abstract: Fault diagnosis of battery power system can clear the fault type, locate the fault location, avoid the failure, and it has very positive effect to increase the stability of electric cars. According to the statistical analysis of electric car big data, this paper researches the evolution regulation and abnormal changes of battery voltage, which accordingly determine the probability of battery fault. Finally, corresponding to the actual vehicle, the statistical fault diagnosis conclusions convert into actual vehicle fault diagnosis conclusions. According to the statistical analysis methods of big data, this paper applies 3σ multi-level screening fault diagnosis which based on Gaussian distribution on determining the fault probability of the battery cell terminal voltage. For the fault statistical analysis of large numbers of electric cars, neural network is used to model big sample statistical law and fit. Applying the neural network algorithm, this paper combines the single car's fault diagnosis results with big sample statistical regulation, construct a more complete battery system fault diagnosis method, and make a corresponding analysis between the statistical result and actual vehicle.