Design And Realization Of Early Warning System For Rolling Bearing Failure Of CNC Machine Tool

With the rapid development of industry and technology,machine tools are becoming more and more intelligent.The fault diagnosis technology of equipment components has matured,but fault diagnosis can only classify the faults of components,and cannot intelligently warn.The rolling bearing fault early warning system is favored by everyone.The system has the ability of status detection,predicting faults and providing early warning suggestions.However,most of the previous systems provide warning indirectly through the failure mechanism of the bearing and its service performance.Therefore,this paper uses deep learning technology to predict the remaining life of the bearing,gives early warning measures through the results of the life,and combines software engineering theory to build an easy-to-operate early warning system.First,the structure of the rolling bearing is analyzed,and the degradation type of the bearing is given.Introduce the causes of bearing vibration.By analyzing the vibration frequency of the bearing,compare the difference between the vibration caused by its own operation and the vibration caused by degradation,and provide a theoretical basis for the vibration signal to be used to predict the life.Summarize the degradation law of bearing,and give the relationship between bearing remaining life prediction and failure warning.At the same time,a new improved mean square error is proposed as the loss function of the network,and good results are achieved.Through predictive analysis of the test data of the bearing life prediction experiment,this method can effectively predict the remaining life of the bearing.Then,for the situation that the remaining life of rolling bearings is difficult to predict,based on the analysis of the characteristics of bearing original signal,it is difficult to extract the signal preprocessing of noise reduction autoencoder and the prediction method of remaining life of bearings based on multi-scale convolutional neural network.In this method,the original vibration acceleration signal is preprocessed by a noise reduction autoencoder,and then the pre-processing result is used as input,and then processed through four parts: shallow feature extraction module,deep feature extraction module,data fusion module,and output module.Finally,the predicted remaining life is output.At the same time,a new type of improved mean square error is proposed as the loss function of the network,which has achieved good results.This method can effectively predict the remaining life of the bearing by predicting and analyzing the test data of the bearing life prediction experiment.Finally,the overall planning of the bearing failure early warning system was carried out,and the system was provided with services by designing software and hardware.The hardware part selects the accelerometer model LIS3 DSH to collect bearing vibration signals,caches the Redis analog message queue and the InfluxDB database to persist the vibration signals.The software part generates the underlying network model through the Keras framework,uses each module of the Java development system,and adopts a front-end and back-end separated architecture to realize user login and management,device management,permission control,data input,life analysis,and early warning.