The Research On Fault Diagnosis Method Of Rolling Bearings Based On Mode Decomposition

Rolling bearings are extremely susceptible to failure damage due to long-term operation in high temperature,heavy load and other harsh environments.Vibration analysis is the typical method to achieve rolling bearing fault diagnosis.However,with the influence of speed fluctuations and background noise,rolling bearing fault vibration signals are often highly nonlinear and non-stationary.Mode decomposition method is one of the powerful tools for signal processing,which extracts the modes associated with the fault information,and removes the excess noise from the signal effectively.Therefore,this paper investigates the fault diagnosis of rolling bearings under different speed conditions on the theoretical basis of mode decomposition method.The main research of the dissertation is as follows:(1)Variational Mode Decomposition(VMD)is introduced in order to extract the resonance frequency bands of rolling bearings and remove the excess noise within the signal.Additionally,a rolling bearing fault diagnosis method based on VMD and modified scale-space representation is proposed.The bearing signal spectrum is divided by modified scale-space representation,and the ratio of fault characteristic amplitude is introduced as a quantitative index to determine the optimal decomposition parameters of VMD.Simulation and experimental results demonstrate that the proposed method is capable of designing suitable bandpass filters adaptively and has certain advantages in extracting resonance frequency bands.(2)The mode components associated with bearing failures under variable speed conditions exhibit broadband and time-varying characteristics generally.Adaptive Chirp Mode Decomposition(ACMD)can effectively extract the broadband time-varying components,yet the reconstructed accuracy of the components is relatively poor under the interference of the proximity components.For this issue,Joint Nonlinear Chirp Mode Decomposition(JNCMD)is proposed.Joint extraction framework is adopted by JNCMD to extract all mode components simultaneously.Through numerical simulation and bearing simulation signal verification,it is demonstrated that the noise robustness of JNCMD and its application potential in broadband time-varying component extraction.(3)Before decomposing the signal by JNCMD,it is necessary to set the number of decomposed modes and the initial instantaneous frequency of each mode.As the rolling bearing failure vibration signal has a complex composition,the input parameters are difficult to determine directly.For this issue,a rolling bearing fault diagnosis method based on fault characteristic order spectrum-guided JNCMD is proposed.The bearing fault signal is resampled from the time domain to the angular domain by angular resampling technique,and the two input parameters of JNCMD are determined according to the fault characteristic order spectrum.By this means,mode components associated with bearing failure can be extracted.Through the comparative analysis of experimental signals,the proposed method can extract the single mode component related to the fault characteristics accurately and improve the time-frequency resolution of the time-frequency representation effectively.