基于EWT多尺度模糊熵和PSO-ELM列车轮损伤检测

Damage Detection of Train Wheel Based on EWT Multiscale Fuzzy Entropy and PSO-ELM

针对车轮损伤信号难以检测的问题,提出基于经验小波变换(EWT)的多尺度模糊熵和粒子群优化极限学习机(PSO-ELM)的车轮损伤检测方法。采用EWT分解信号获得若干模态分量,根据相关系数和欧式距离指标,选择最优模态分量,求取其多尺度模糊熵作为特征向量,输入到PSO-ELM分类器中进行分类识别,并与PSO-LSSVM进行对比。研究结果表明,该方法能够有效检测车轮损伤状态,其识别准确率高,效率高,具有较强的工程实用价值。

Aiming at the difficulty of detection of signals of the damaged wheel,a detected method of the damaged wheel based on empirical wavelet transform(EWT)multi-scale fuzzy entropy and particle swarm optimization extreme learning machine(PSO-ELM)is proposed.Firstly,some modal components are obtained by EWT decomposition.Then,according to the correlation coefficient and Euclidean distance index,the optimal modal component is selected,and its multi-scale fuzzy entropy is obtained as eigenvector.Finally,the eigenvector is input into the PSO-ELM classifier for classification and recognition.Compared with PSO-LSSVM classifier,The results show that this method can effectively detect wheel damage state,and has high recognition accuracy,high efficiency and strong engineering practical value.