Study On Gearbox Fault Diagnosis Technology Based Information Fusion

In the producing industry, in order to make good use of the equipment, people alwayshope to predict the equipment fault. Along with equipment's development towards high-speed,high-power and high-stabilization, it is difficult for the traditional method to diagnose thecomplex equipment fault. In view of this, we introduce the technique of information fusioninto the field of Gearbox fault diagnosis.This article is mainly about Gearbox fault diagnostic method. Based o analyzing thecurrently domestic and foreign study, it gives a new diagnostic method based on neural netand D-S inference, and it is verified.Firstly, Based on studying the traditional diagnostic methods and their characteristic, thestudy and application of new method is urgent required.Secondly, we discuss the network structure and the characteristic of BP neural network,and it shows that neural net can be used for Gearbox fault diagnosis. We study Gearbox faultdiagnosis means based BP neural network, put forward and upswing method integratingalter-learn rate and appending momentum, which quicken the convergence speed, and theresult based on the simulation of the example show it is a effective way by using network todeal with the characteristic information. To the incertitude of fault diagnosis, thedecision-making fusion based on D-S evidence theory is put forward. The basic conception ofevidence theory is introduced , and with the simulation based on the example, we knowevidence theory fusion improved the fault diagnosis precision.Finally, the method based on BP neural network and D-S inference is put forward. Bythe simulation of the example , the feasibility and availability of this diagnostic method areverified, it can increase the accuracy of fault diagnosis.