Based On Model Analysis And Informational Entropy In Gearbox Failure Diagnosis Research

In nowadays, gearbox is a component of energy transmission, which cannot beneglected in mechanical structure. Therefore, the research of gearbox failure diagnosis isgreatly significant. In this article, the typical failure form, vibration mechanism and thecharacteristic of gearbox vibration signal had been analyzed. To be specific, this researchconsider the three types of model analysis methods, computational model analysis,experimental model analysis and operational model analysis, while using the operationalmodel analysis to study gearbox failure diagnosis. To the question of gearbox vibrationsignal in operational model analysis, the informational entropy theory is introduced,meanwhile, the entropy calculation which based on cross-power spectrum has beenpresented for comparison and identification to achieve the goal of failure diagnosis.Firstly, the research need to establish the JZQ-250gearbox experimental platform,select experimental instruments by requests, set various failures and make experimentalplan. Through using UG and ANSYS, the gearbox can be analyzed by solid modeling andfinite element analysis. In addition, the PolyMAX can be used to conduct modal parameteridentification to frequency-response function of gearbox, in order to obtain the experimentalmodel analysis result. Combine with the results and test point arrangement, basing on modalvibration type to set the experimental test point to gather the vibration signal of constantrotational speed gearbox and speed-increasing gearbox.Secondly, researching gearbox operational model analysis method, using variousreferenced cross power spectral matrices instead of matrix of frequency response function,identifying system pole by using Op.PolyMAX method. This method can be used to analyzethe uniform velocity and variable velocity vibration signals of normal and faulted gearbox,while identifying the model vibration frequency and shape to compare with the results ofoperational model analysis, in order to find out the difference and realize failure extraction. Especially, the harmonic component of gearbox vibration signal is analyzed and rejectedduring the research of variable velocity vibration signal analysis.Finally, researching the informational entropy theory, presenting the cross powerspectral entropy calculation, and calculating the entropy value of SUM cross powerspectrum in different operational environment. Combining with singular spectrum entropyand power spectrum entropy of vibration signal to identify by neural networks forachieving the diagnostic purposes.