基于小波包特征熵的高速列车监测数据的特征分析

Characteristics Analysis of High-speed Train Monitoring Data Based on Wavelet Packet Characteristic Entropy

将高速列车在不同工况和速度下的监测数据进行傅里叶分析用来确定各种不同工况信号的频率范围,再对不同工况和速度下的信号进行小波包分解,并重构通频范围内前几个低频带信号,进而建立信号的小波包特征熵向量,不同频带信号的小波包特征熵变化反映了列车运行状态的改变,最后将得到的小波包特征熵向量输入支持向量机进行故障识别。仿真分析结果表明该方法对高速列车故障状态识别是有效、可行的。

Firstly Fourier analysis was used to determine the frequency range of the highspeed train monitoring data under different conditions and the speed.Then the signal under different conditions and speed in the first few low frequency band were decomposed and reconstructed by wavelet.thereby the wavelet packet characteristic entropy were established.The changes of wavelet packet characteristic entropy in different frequency bands reflect a change in the status of the train running.Finally,wavelet packet characteristic entropy vectors were input in SVM（support vector machine）for detecting fault.After simulation analysis of experimental data,the results show that the method for high-speed train fault state identification is valid and feasible.