Research And Design Of Rolling Bearing Fault Diagnosis System In Cement Production Line

Rolling bearing is an important component of the key equipment of the cement production line.Its working environment is bad and its working intensity is high.It is the location of frequent failures.Once a failure occurs,it will affect the production of the whole cement production line,produce potential safety hazards,and cause economic losses to the cement plant.It is of great significance to detect the failure to ensure the safe operation of the system.At present,the fault diagnosis methods of rolling bearing can be divided into two kinds according to the angle of signal processing,one is to rely on signal processing for diagnosis,the other is to apply the deep learning which has received great attention in recent years to fault diagnosis.The core of the first diagnosis method is to gradually find out the fault characteristics of the signal through a series of processing of the collected fault signals,and finally achieve the purpose of signal classification;the core idea of the second diagnosis method is to apply the relevant knowledge of deep learning,in the background of the obtained signals showing massive characteristics,the first diagnosis method will exist greatly The limitations of deep learning,which can achieve the purpose of extracting features through its internal network without additional processing of the signal,so as to achieve the classification of the signal.Compared with the traditional manual detection of fault diagnosis,these two methods are more simple in application,and can be detected in the stage of slight fault,which is not obvious,and the diagnosis rate of fault is higher.Therefore,this paper will study the method of rolling bearing detection under these two backgrounds.First of all,this paper studies the signal fault diagnosis method of rolling bearing.On the basis of signal processing technology,this paper puts forward the processing method of combining wavelet decomposition and reconstruction technology with spectral kurtosis,studies the method of using wavelet packet to decompose and reconstruct the signal,obtains the kurtosis graph of the signal,designs band-pass filter to extract the impact frequency band of the signal,and finally according to the principle of The envelope spectrum extracts the fault characteristic frequency of the signal to determine the fault type of the signal finally.After the algorithm is proposed,this paper uses simulation data and test-bed data to verify the algorithm on the basis of experiments.Then,this paper also studies how to apply DBN to the fault diagnosis of rolling bearing.In the deep learning,the determination of data set plays an important role.After that,it analyzes the relevant parameters of RBM that affect the accuracy of DBN model,and determines the number of iterations of RBM,the learning rate and the number of layers of RBM in DBN network for the diagnosis of the whole model The final DBN structure network is determined by the influence of break rate.The model has a good classification effect after the test-bed data verification.Finally,on the basis of these two fault diagnosis methods,this paper designs a research application platform for rolling bearing fault diagnosis under the research environment.The platform takes MFC as the main development framework and SQL as the main development framework Server database technology,matlab technology,can be used to analyze and process the collected sample data in the laboratory environment,realize the data diagnosis,at the same time,it can intuitively compare the effect of various fault diagnosis algorithms,which is convenient for researchers to compare the performance of various algorithms.Through the experiment test,the system runs stably and has a high rate of fault identification,which has a high value of scientific research and application.