Research On The Method Of EMD And Neural Network Of Particle Swarm Optimization In The Fault Diagnosis Of Gearbox

This paper based on the background of gearbox fault diagnosis. In this paper, a newlyself-adaptive time-frequency analysis method in recent yearsis mainly studied, which is theempirical mode decomposition (EMD). It can effectively apply to non-stationary signal andcan get significative instantaneous frequency, thus provides a new method for the diagnosis ofmechanical equipment.In this paper , the basic theory and realizing process of EMD is researched. According tothe actual application problems of EMD, the paper proposed the improved method of EMD.Then the paper analyzed the end effect of EMD, and tried the method of extension of extremepoint to solve the end point effect.In the second part, the type of JZQ250 gearbox in the laboratory was made as test object.After that the paper applied the method of improved EMD to the field of fault diagnosis ofgearbox, and based on the fault mechanism and vibration trait of the gearbox, the vibrationsignals of the gearbox were tested and the method of improved EMD was applied to analyzethe signal and to diagnose the fault of snaggletooth, bearing cup and bearing cone, and got agood diagnosis effect. The test results proved the availability of EMD in gearbox faultdiagnosis field. Finally, the faults of gear-box are diagnosed by trained PSO-RBF neuralnetwork, enhanced the precision for gearbox fault diagnosis and had important significancefor drive the develop of gearbox diagnosis technology.