ALIF Method And Its Application To Rolling Bearing Diagnosis

A key step in the diagnosis of rolling bearing faults is how to extract effective fault signature information from complex rolling bearing fault signals.Traditional signal processing methods such as Empirical Mode Decomposition(EMD)and Local Mean Decomposition(LMD)are local features that depend on the extreme point distribution of the original signal and can be adaptive.The rolling bearing vibration signal is decomposed into several physical meaning components to complete the fault diagnosis of the rolling bearing.However,there are some shortcomings.For example,the model mixing,the boundary effect and the noise resistance is not strong.These all reduce the reliability of rolling bearing fault diagnosis to some extent.Therefore,the introduction of a new signal adaptive time-frequency analysis method is very important for the research field of rolling bearing fault diagnosis.Based on this paper,the basic principle of Adaptive Local Iterative Filtering For Signal Decomposition(ALIF)and its application in fault diagnosis of rolling bearings are studied.The main research contents related to the thesis are as follows:(1)The basic principle of the ALIF method is studied.The simulation signals are compared with the EMD,LMD and LCD method result.It is demonstrated that the ALIF method has the same decomposition ability as the EMD method.At the same time,the ALIF method is applied to the fault diagnosis of the outer ring of rolling bearings.This demonstrates the feasibility of applying this method in the field of rolling bearing fault diagnosis.(2)For the model mixing problem that occurs in the decomposition results of the ALIF method,Complete Ensemble Adaptive Local Iterative Filtering Decomposition(CEALIF)method based on noise-assisted analysis is proposed.The improved CEALIF method is applied to the simulation signal and compared with EEMD and CEEMDAN method.The ability of the CEALIF method to suppress the model mixing problem is studied.At the same time,the CEALIF method is successfully applied to the gearbox composite fault.(3)For the boundary effect of ALIF decomposition results,three data extension methods,namely image extension and SVR data extension are proposed,and the data extension method is adopted.The ALIF decomposition results are compared with those before the extension,the feasibility and practicability of the data extension method in solving the boundary effect problem of the ALIF method is demonstrate.(4)IALIF decomposition method is proposed by analyzing the results of the research on ALIF decomposition method.The ability of IALIF to suppress model mixing and boundary effect is demonstrated by comparing simulated signals with EEMD decomposition results using SVR data extension.For the rolling bearing signals obtained in actual working conditions,which often contain a lot of noise and other redundant information,Fault diagnosis method of rolling bearing combining IALIF and CNN is proposed.The basic idea is to first use the IALIF method to decompose the rolling bearing fault signal to extract the effective feature component,and then input the data set into the CNN model to complete the training.After the training is completed,the rolling bearing pattern recognition can be completed.The experimental analysis of the actual rolling bearing signal demonstrates that the method can complete the qualitative diagnosis of rolling bearing faults,and the recognition rate is high,which has certain practicability.