Research On Rolling Bearing Fault Diagnosis Method Based On MCKD And CN

Nowadays,the development of modern industry makes machinery and equipment develop more and more in the direction of intelligence and precision.Mechanical equipment is applied in various fields,and its development provides convenience for people’s production and life.The rolling bearing is one of the key components of mechanical equipment,and its working state affects the working state of the entire equipment.Therefore,it is of great significance for industrial production to select a reliable fault diagnosis scheme for periodic condition monitoring of rolling bearings.In actual industrial production,the operating conditions of the bearing are very complex,the measurement signal is affected by the background noise,and compound faults occur from time to time,which makes the vibration signal complex and difficult to distinguish,and the fault features are coupled with each other,which increases the difficulty of fault diagnosis and analysis of the bearing.For the traditional fault diagnosis methods,it is difficult to extract effective fault features from a large number of fault data sets,which largely depends on experience and knowledge.Therefore,a rolling bearing fault diagnosis technology based on parameter optimization maximum correlation kurtosis deconvolution(MCKD)and convolution neural network(CNN)is proposed in this paper.The signal preprocessing method is applied to the filtering analysis of vibration signals,and a convolutional neural network model is constructed to complete the single and compound fault diagnosis of rolling bearings to analyze and explore the effectiveness of the proposed method.The main research contents are as follows:(1)Aiming at the characteristics that the bearing vibration signal is easily affected by various factors such as background noise and simultaneous occurrence of various faults,a signal preprocessing model of parameter optimization MCKD is established.Through the comparison between particle swarm optimization algorithm(PSO)and cuckoo optimization algorithm(CS)from the aspects of theoretical basis and experimental verification,this paper chooses the more novel and better optimization performance of the adaptive multi-strategy cuckoo algorithm(MSACS).The important parameters of MCKD are optimized by MSACS adaptive iteration.The optimized MCKD algorithm performs signal filtering and noise reduction on the rolling bearing fault signal,and obtains the signal after noise reduction.(2)Aiming at the problem that the traditional bearing fault diagnosis method relies heavily on experience and knowledge,a fault diagnosis model combining parameter optimization MCKD and CNN is established.The filtered signal is input into the constructed CNN model for training and testing,and get the classification result of rolling bearing fault diagnosis.The feature distribution is visualized and analyzed by the data dimension reduction algorithm t-SNE technology.Finally,for the model methods proposed above,the fault vibration data provided by Case Western Reserve University and the vibration data collected by the bearing life test bench are used for experimental verification,and the proposed method is compared with other similar model methods to prove the effectiveness of the proposed diagnosis method.