Research On Prediction Model Of Gear Wear Based On On-line Oil Particle Monitoring

As an important part of high-end equipment in aerospace,transportation,power and petrochemical fields,mechanical power transmission system plays an important role in the field of national defense and national economy.Once it fails,it will lose a lot.As the most critical part of power transmission system,gear wear status will directly affect the reliability and safety of mechanical system.In engineering practice,people expect to know the wear status and wear trend of gears well,so as to prevent uncontrollable conditions from occurring.On-line monitoring technology of oil particles can realize real-time,continuous and on-line monitoring of wear particles in oil.this technology is turning traditional preventive maintenance into predictive maintenance and promoting the development of mechanical maintenance system towards intelligent direction.Therefore,it has important academic and engineering application value to study how to apply the On-line monitoring technology of oil particles to monitor wear status and predict wear condition of gear.In this paper,we carry out research on prediction model of gear wear based on on-line oil particle monitoring.The main contents of this paper are as follows:Firstly,test system of on-line oil particles monitoring is established.According to the requirement of wear test and the MCL-1 closed power flow test-rig,the design scheme of oil circuit system is put forward.online oil particles monitoring sensor ZXA-07 and oil return pump are installed in the return oil circuit of gearbox.Two-stage filter system is added in the oil supply circuit to realize the coarse filter before pump and the fine filter after pump.Auxiliary installation for connecting the filter system and the online oil particles monitoring sensor is added.The remote monitoring and control system is established to realize the remote control of the running state of the equipment.Secondly,test research of on-line oil particles monitoring is carried out.According to the purpose of the test,the test scheme is designed.The parameters of the test gear and the test lubricating oil are selected.the test load and test speed are determined.The gear wear test is carried out.The test data are analyzed and the variation rule of the characteristic parameters of the lubricant abrasive particles with the gear wear is analyzed.Thirdly,a prediction model of wavelet neural network based on genetic annealing optimization(GA-WNN)is proposed.firstly,the wavelet neural network is established by replacing the hidden layer neurons of BP neural network with wavelet elements.secondly,GA algorithm is proposed to optimize the initial weights,thresholds,scaling factors and translation factors of the wavelet neural network.finally,the optimized initial parameters of the network are used to train the network,and the prediction is obtained.Finally,gear wear prediction research based on on-line wear particles monitoring is carried out.The forecasting model proposed in the previous chapter is validated by an application example.the data cleaning method commonly used in data mining is used to clean the wear characteristic data monitored by sensors.the time series-based gear wear prediction model is established by GA-WNN to predict the gear wear trend.