Study On Order Analysis-based Fault Diagnosis Method Of Rotating Machinery

Process of speeding and slowing down contains rich fault information in rotatingmachinery. Order analysis can make full use of this part of information. As a result, it isone of the most important techniques of rotating machinery to deal with vibration signal.At present, there are some special characteristic analyzers of rotating machinery on themarket. Although their precision is high, certain specific hardware must be used, such astachometer, photoelectric encoder and order tracking filter. It is expensive and is notconvenient to be used. That limits its extensive use in rotating machinery vibration test.So how to use software to process signal with order analysis and ensure the accuracy is aresearch focus in this article.This paper studies deeply order analysis and various time-frequency analysismethods. Aiming at the defects of existing methods, the wavelet ridge line-based orderanalysis method is presented. Wavelet ridge line is the meeting of modulus maximapoints of wavelet coefficient at each moment in time-frequency plane. It can clearlycharacterize the change of instantaneous frequency. The instantaneous frequency ofsimulation signal is extracted respectively by using wavelet ridge line method and Hilberttransform. Those two ways are compared to tell others which is better.In addition, EEMD method proves its advantage in eliminating the modal aliasingexisting in EMD by simulation. An approach based on the theory of spectral kurtosis andEEMD is used to extract the IMF containing fault sensitive components. It combinesspectral kurtosis with energy spectrum and is applied to deal with the fault signal inexperiments.In order to verify the effectiveness of this method, hydraulic system of vibration andspeed test bench is set up. Vibration signal and speed signal of pump are collected at thesame time and are researched. With wavelet ridge line-based order analysis technology,speed of motor shaft is tracked and fault characteristics order is extracted accurately. In aword, order analysis can be achieved in software with high precision. In combinationwith K-means algorithm, fault diagnosis of hydraulic pump goes successfully.