Research On Small Fault Diagnosis Of Rolling Bearing Based On LMD And Chaotic Fractal

Rolling bearing is one of the most important and the most prone to failure components in rotating machinery. The bearing fault directly affect the proper functioning of equipment, even cause serious accidents, so it is very important to have a careful diagnosis in the rolling bearing. It is the key to bearing fault diagnosis that is the fault feature extraction. However, duo to the influence of transmission path and noise, the process of rolling bearings is a dynamic process which is complex and non-stationary, and whose vibration signal also exhibits strong nonlinear and non-stationary, greatly increase the difficulty of the rolling bearing fault diagnosis, especially slight fault diagnosis. The fractal and chaos theory has a good representation ability for nonlinear signal, it is widely used in the feature extraction of nonlinear digital signals. Based on that, the chaotic fractal theory and local mean decomposition (LMD) are introduced to analyze the vibration signal of rolling bearings in both time domain and time-frequency domain, to identify its running status. Thesis research work are as follows:(1)Chaotic characteristic analysis of rolling bearing signal is based on the largest Lyapunov exponent. Before analyze the vibration signal by fractal theory, its chaotic characteristics are identified by the largest Lyapunov exponent, to verify the feasibility of fractal theory for bearing vibration signals.(2)Rolling bearing fault diagnosis based on fractal dimension. The correlation dimension is used to identify the bearing vibration signals in time domain, which with different state and different fault degree (slight faults and obvious faults). For the signal which cannot be distinguished by correlation dimension, the box dimension will be introduced to confirm the recognition results, and identify the fault of rolling bearing.(3)Rolling bearing slight fault diagnosis based on LMD and fractal box dimension. Extract the fractal feature of bearing vibration signal in time-frequency domain. Decompose the original vibration signal by LMD, and calculate the time-frequency distribution of useful PFs. Further, the box dimension of time-frequency domain is obtained and take it as the characteristic parameter for slight fault diagnosis of rolling bearing.(4)Rolling bearing slight fault diagnosis method which is based on improved LMD and multifractal dimension. Aiming at the problem of mode mixing effect in LMD decomposition, the method of removing the mode mixing by mask signal is introduced; and the multifractal characteristics of vibration signal in the time-frequency domain is extracted, which is used as the fault feature for the slight fault diagnosis. The experiment of rolling bearing data shows that multifractal dimension based on improved LMD can identify the slight fault of bearing more accurately.