Abstract: To avoid the potential safety hazards of electric vehicle caused by fault deterioration of in-wheel motor, a condition recognition method based on artificial hydrocarbon networks with K-means clustering algorithm(K-AHNs) is proposed. Based on the K-means clustering algorithm, the traditional updating method is improved to optimize the molecular interval and multiple condition recognition models of artificial hydrocarbon networks(AHNs), then the fault recognition accuracy is improved and the training time is significantly reduced. The experimental data of 3 bearing faults such as inner-ring fault, outer-ring fault and rolling element fault of in-wheel motor with 4 kinds of roads in 7 running conditions are used to verify the performance of the proposed method. Each condition recognition accuracy of the proposed method is greater than 87%, and each training time is less than 19 seconds. Comparing the traditional AHNs, the average recognition accuracy based on K-AHNs is increased by 14.49%, and the average training time is 5 times shorter. It has been proved that K-AHNs not only improves the condition recognition accuracy, but also reduces the training time, then K-AHNs has better reliability and practicability, which solves some problems of the existing fault diagnosis of in-wheel motor, such as low recognition accuracy and long training time.

Key words: in-wheel motor, artificial hydrocarbon networks, K-means clustering, condition recognition, running condition