Gearbox fault diagnosis based on LMD and cyclostationary demodulation

Abstract

A ZS-65 gearbox system is as the study object, the gear's characteristic frequency and bearing's characteristic frequency of gearbox are determined based on analysis of the test platform, structure parameters, operating principle, and the layout of measuring point and vibration signals are collected in normal status, tooth wear, and outer ring failure condition, the motor speed is set as 1200r/min, the sampling frequency is 4000Hz, within 1s samples signal are selected and analyzed. The vibration signals are directly demodulated by using cyclostationary demodulation method, and the cyclic spectrum and cycle frequency axis projection are got, and the frequency aliasing phenomena are analyzed. On this basis, by using the local mean decomposition which is based on the signal itself information adaptive decomposition, and resulted PF component has a real physical significance, obtained time-frequency distribution can be clearly and accurately reflect the distribution regularities of signal energy in space various scales. Based on the vibration signal, first LMD decomposition downs to it as individual components' modulated signal, the energy of various components are calculated in addition to the residual value, and normalization processing, wherein the decomposed energy concentrated product functions are proceeded the cyclostationary demodulation, obtained their cyclic spectrum, then the cyclic spectrum are combined and extracted and analyzed slices map in mesh frequency, it inhibits the influence of cross interference items directly uses cyclostationary demodulation, and gets a good result.