Optimization Study On Multi-Parameter Fault Diagnosis Method Of Gearbox Based On Vibration Monitoring And Oil Analysis

Compared with the single-parameter fault diagnosis method,the multi-parameter fault diagnosis method can more accurately diagnose the fault condition of the equipment.However,in practical applications,the development of multi-parameter monitoring programs is susceptible to the knowledge level and experience of practitioners.The monitoring scheme with single sensitive objects will lead to insufficient diagnostic accuracy,and the monitoring scheme with redundant project functions will lead to high monitoring cost.In order to explore a method for formulation multi-parameter monitoring scheme,this paper takes the gearbox as the research object,and explores the multi-parameter fault diagnosis method based on vibration monitoring and oil analysis.The main contents are as follows:Firstly,on the basis of exploring the internal connection of gear tooth surface failure mechanism with its vibration signal and oil information,the time-domain waveform and frequency-domain spectrum of vibration signal,physical and chemical performance indicators and abrasive particle information of oil,and other common equipment condition monitoring items are analyzed,the multi-parameter condition monitoring scheme of the gearbox to be optimized are determined.It provides theoretical support for the optimization research of fault diagnosis methods.Secondly,in order to obtain accurate information of gearbox wear fault condition,the gearbox fault simulation experiment was carried out on the QPZZ-II rotary machinery fault simulation test rig.On the basis of the characteristics of gear wear failure conditions,the whole life cycle fault condition is according to the five stages of normal,slight wear,abnormal wear,severe wear and failure,simulation method of the gear is designed and realized.A vibration monitoring system consisting of vibration sensor,acquisition instrument and signal analysis software is constructed to collect gearbox fault vibration signals and extract relevant characteristic parameters.An oil analysis system consisting of a capillary viscometer,a particle counter and a rotary ferrography analysis system is constructed to collect gear oil samples and extract relevant characteristic parameters.Then,in order to realize the multi-parameter fault diagnosis method of gearbox based on vibration monitoring and oil analysis,on the basis of the principle of solving the fault diagnosis problem of neural network,the fault diagnosis model based on BP neural network is constructed.The training sample set of the fault diagnosis model is established,which based on gearbox failure simulation test result.The simulation and training of the gearbox multi-parameter fault diagnosis basic model are completed,and the training results of the model are summarized.Finally,aiming at the problem of prone to overfitting of gearbox multi-parameter fault diagnosis basic model,an optimization method based on genetic algorithm is proposed.On the basis of completing the algorithm framework and related operator design,the fault diagnostic basic model is optimized,the redundant items in the multi-parameter monitoring scheme are eliminated,and the optimal initial connection weight and threshold of the fault diagnosis model are determined.Through the model performance comparison test,the correctness and effectiveness of the optimization method for multi-parameter fault diagnosis model of gearbox based on genetic algorithm are verified.