Fault diagnosis of gear based on singular value decomposition and RBF neural network

Aiming at the problem that the weak gear fault signal is difficult to detect, a gear fault diagnosis method based on singular value decomposition and RBF neural network is proposed. Firstly, the four failure signals of normal, broken tooth, crack and wear were collected through the experimental table, then the signals were denoised by the singular value decomposition. After the denoising signals were decomposed by the three-layer wavelet packet, the energy eigenvalues of eight frequency bands were calculated and the energy eigenvalues were used as input samples of RBF neural network training. Through the experimental data of the sample test, the results show that the proposed method has the advantage of high diagnostic accuracy and fast diagnosis time.

I. Introduction

All Gear is the main component of rotating machinery and equipment, and also prone to failure of parts. Most of rotating machinery need to use the gears in the general industrial, which occupy a high position in the mechanical equipment, academia and industry are very concerned about the gear running state monitoring [1], [2]. Gear failure may lead to equipment noise and abnormal vibration, Severe failures can cause some severe problems that the entire plant or even the entire production line cannot work properly.