基于WE-ICA的牵引电机速度传感器微小故障检测与识别

Incipient Fault Detection and Identification of Traction Motor Speed Sensor Based on WE-ICA

文章提出了一种基于小波变换和集成独立成分分析(WE-ICA)的牵引电机速度传感器微小故障检测与识别方法。首先,通过小波变换对高速列车牵引电机数据进行滤波处理,减少噪声对微小故障信息的干扰;接着通过独立成分分析(ICA)方法提取故障信息并建立检测统计量;最后利用贝叶斯推理计算统计量的故障概率,并设计集成统计量用于微小故障检测;另外通过加权贡献度对故障进行识别。利用CRH2牵引电机实验平台对该方法进行验证,实验结果证明了该方法的有效性和实用性。

It presented a method for detecting and identifying incipient faults in traction motor speed sensor of high-speed train based on wavelet transform and ensemble independent component analysis (WE-ICA). Firstly, the data of traction motor of high-speed train are filtered by wavelet transform to reduce the interference of noise to the information of incipient faults. Secondly, the fault information is extracted by independent component analysis (ICA) and the detection statistics are established. Finally, the probability of system faults is calculated by Bayesian reasoning and the ensemble statistics are designed. In addition, weighted contribution plot is used to identify the fault. The proposed method was validated by CRH2 traction motor experimental platform. Experimental results show the effectiveness and practicability of the proposed method.