Gearbox Fault Diagnosis Based On Null Space Pursuit And Marginal Time Spectrum Of S Transform

As a key technology to protect the normal operation, fault diagnosis of mechanicalequipment has all the time attracted broad attention. Further, the gearbox is the powertransmission device of the whole mechanical system, which directly affects the quality of theoperating state of the mechanical system. Therefore gearbox fault diagnosis technology is ofimportant engineering value and factual significance.In this paper, research of gearbox fault diagnosis technology is presented based onvibration signal analysis. Null Space Pursuit algorithm which has come into being in therecent years is adept at the decomposition of non-stationary signals and denoising. Combinedwith Null Space Pursuit algorithm, marginal time spectrum of S transform is proposed todiagnose gearbox fault and achieved good results.Firstly, current status of the study and development on gearbox fault diagnosis isdescribed in detail as well as the popular fault diagnosis technology. Besides, we alsointroduce the mechanism of gearbox fault, the main failure types, signal characteristics andcommon signal processing methods. As for Null Space Pursuit, theoretical basis andcalculation process is elaborated for subsequent application. Meanwhile, the advantage ofNSP for decomposing signal, processing non-stationary signal and denoising was highlightedwhen did simulation and comparison with EMD method, the simulation focus onnon-stationary signal with noise. Marginal time spectrum of S transform was proposed stepby step through processing actual bearing fault signal; it has a good performance inprocessing weak fault signal. Furthermore, a hybrid approach combing Null Space Pursuit andmarginal time spectrum of S transform was proposed to diagnose gearbox fault whichincludes wear of tooth surface, cage fault, rolling element fault and outer ring wear. Theexperiment results show that the hybrid approach can successfully reflect the time-frequencyspectrum variation and get the fault characteristic frequency which accord with theoreticalresults. Fault diagnosis of rolling success also shows that the method is effective for weakfault diagnosis...