Research On Minimum Entropy Deconvolution For Rolling Element Bearing Fault Diagnosis

With the rapid development of technology and industry, mechanical equipment hasbecome more and more huge, complex, high-speed, effective, and heavy-load while they mustface more and more harsh running conditions. Once they fail unexpectedly, the unexpectedfailure can increase maintenance cost, reduce production efficiency, and sometimes causesignificant economic losses, or even catastrophic accidents. Therefore, it is necessary and ofgreat significance to carry on new fault diagnosis technology research.How to effectively extract the true signal or the signal feature under strong noise andapply to practical engineering is the core part of mechanical equipment condition monitoringand fault diagnosis. And it is also a hot research field of fault diagnosis. New methods andtheory come out one after another to enrich and improve the mechanical fault diagnosistechnology. Rolling element bearings which are common in rotating machines are the researchobjects of the paper. The feasibility and validity of utilizing the minimum entropydeconvolution in weak fault diagnosis and condition monitoring is demonstrated by bothsimulation and experiments. The paper mainly including the following aspects:(1) From the viewpoint of theoretical analysis and engineering application, this paper’sresearch background and significance of present study are elucidated. A state of the art reviewis thoroughly completed, which consists of fault diagnosis of machinery and equipment, therolling bearing fault diagnosis, time-frequency analysis and information entropy theory. Theissues to be resolved are summarized and the research content of this paper are established.(2) Minimum entropy deconvolution and its related theory are described. First of all,entropy and information entropy is introduced, and its nature of reflecting the signalcomplexity and uncertainty is discussed. Then, the application of entropy in signal processing,mainly spectrum entropy, is introduced. Finally, the theory and the realization of minimumentropy deconvolution are described.(3) Simulation analysis is used to demonstrate the feasibility of utilizing the minimumentropy deconvolution in rolling element bearing fault diagnosis. First of all, as the researchobjects of this paper, the structure of rolling element bearing, the failure type, and the faultmodel is mainly introduced. Then, based on the fault model, this paper illustrates the principleof utilizing the minimum entropy deconvolution in rolling element bearing fault diagnosis.Then simulation analysis is used to demonstrate this point. Finally, the inflect factors ofminimum entropy deconvolution is studied.(4) Experiment validation and analysis. Two experiments are conducted to demonstratethe feasibility and validity of minimum entropy deconvolution. The first experiment is rollingelement bearing accelerated life test. The data analysis shows the minimum entropydeconvolution is effective in extract the weak fault. From the analysis, MED is found to have excellent ability in bearing early incipient fault diagnosis and condition monitoring. And inapplication, fault diagnosis may be very difficult because of bearings are often associated withother equipment such as gears, and the fault information is corrupted by strong backgroundnoise. Therefore, a single stage gearbox test with one inner race fault bearing is performed tocollect experimental data. The results showed that it can detect and diagnose bearing faultunder strong noise. The feasibility and validity of utilizing the minimum entropydeconvolution under strong noise is demonstrated in this experiment.