Research On Hub Bearing Fault Prediction Based On Combination Model

As one of the most important components of the vehicle,the operational reliability of the wheel bearing is closely related to the safety and comfort of the vehicle.In order to guarantee the security of driving for the vehicle and achieve the predictive maintenance of the hub bearing,here,the wheel hub bearing in heavy-duty truck is regarded as the objective of research.Firstly,combined with its structural characteristics and actual operation characteristics,the study of optimization is carried out on the grey model(1,1)and BP neural network.Meanwhile,combination forecasting model which possesses the optimal ability of data fitting is proposed.According to the actual working environment of wheel hub bearing,the test platform of collective intelligence about wheel hub bearing is established.At the same time,the accelerated fatigue life experimentation is carried out to validate the prediction accuracy of the model for the degradation trend of wheel bearing.Finally,the fault prediction system is designed for wheel bearing.Firstly,the classification of fault prediction methods and the research status at home and abroad are elaborated in detail which is based on the background of the project.According to the structural property and actual operation characteristics of wheel bearing,failure principle of vibration,major failure mode and the approach of property extraction of signal about hub bearings are seriously carried out the analysis.Secondly,the principle of predicted model for grey predicted model(1,1)and BP neural network is in-detail described in this article.Besides,on the basis of retaining the original advantages of each single model,the improved grey predicted model(1,1)can be obtained by means of changing the albino background value,and the GA-BP neural network can also be achieved through adjusting initial weight and threshold.At the same time,using the fault case of bearing,the value of accuracy for each model above can be verified.According to the rule of least square error,a combined forecasting model can be constructed,which will finally enhance predicted accuracy.Thirdly,the test bench of collective intelligence about wheel hub bearing is established.At the same time,the fatigue life test is also carried out for wheel bearing.Then,the sample data can also be achieved through the data acquisition instrument in the test-bed.Furthermore,characteristic parameter of hub bearing can also be obtained.According to the relative criterion of the vibration for the device and the degradation trend of actual fault for the test bearing,the damage degree of the test bearing can be determined quantitatively.The following content,which selecting the RMS value as the time series of sample,need to verify the predicted reliability of each model.Finally,the system of failure prediction for hub bearing can be achieved by means of the Visual Basic and MATLAB software.Besides,a variety of functional units in the predictive system have been developed,such as,user login,fault prediction of hub bearing,the input and output of sample,data query and data statistics.