A Study On Gear Fault Diagnosis Technology Based On Support Vector Machine Optimized By Bee Colony Algorithm

Gear as an important part in the mechanical equipment, is widely used in metal cutting,aerospace industry, national defense construction, electricity, transportation and so on all walks of life, the failure probability is higher, in the event of failure would have caused production downtime, or even a threat to the safety, therefore, on the condition monitoring and fault diagnosis has important practical significance. A Study on Gear Fault Diagnosis Technology Based on Support Vector Machine Optimized by Bee Colony Algorithm is mainly studied in this paper.Firstly, the common failure types and causes of gear is studied, gear fault vibration mechanism is analyzed, several time domain index is studied and introduced the topic using the fault simulation test platform QPZZ- II in this paper.Secondly, the basic principle of artificial colony algorithm is studied, according to its convergence speed is slow, late evolutionary population diversity loss may be excessive, even into the local optimal solution of the problem, introducing chaos initialization and tournament selection strategy, through the four standard function optimization simulation experiment validation chaotic artificial colony algorithm has faster convergence speed and higher precision of optimization; The classification of support vector machine(SVM) method isstudied, chose chaos artificial bee colony algorithm to optimization of model parameters,using the UCI database of Heart, Iris and Wine data simulation, validation chaotic artificial bee colony algorithm optimization of support vector machine has higher classification accuracy. Using the time domain parameter before denoising as features, the method was applied to the actual fault diagnosis of gear, get good fault diagnosis effect; Using the time domain parameter after denoising as features, using the method in this paper to the actual gear fault diagnosis,compared the fault recognition results before and after denoising,verify the capacitance noise ability of the time domain parameters is poor.Finally, the basic S transform and fuzzy entropy theory is studied, aiming at the effects of gear noise in the process of signal acquisition and the complexity and non-stationary signal,this paper puts forward the feature extraction method based on generalized S transform fuzzy entropy, generalized S transform of signal to get the time-frequency matrix, then segment the frequency and calculation of the fuzzy entropy spectrum as a signal feature vector, through the analog signal simulation experiments verify the method can effectively reduce the noise influence. The method was applied to actual gear fault diagnosis, using swarm optimization ofsupport vector machine(SVM) as classifier, get higher classification accuracy.