Research on fault diagnosis of railway point machine based on multi-entropy and support vector machine

Abstract A new fault diagnosis method is proposed to effectively extract the features of sound signal typical faults ZDJ9 railway point machines. multi-entropy feature extraction by combing multi-scale permutation entropy and wavelet packet entropy. Firstly, empirical mode decomposition performed on signals obtain modal components with different time scales. Then, extracted from these components. Meanwhile, sensitive nodes obtained analyzing reconstructed last layer nodes. Since arrangement can distinguish subtle signal, original be as vector machine in states. To reduce redundant information among high-dimensional features, ReliefF utilized. Finally, support (SVM) used judge type machine.