Research On Feature Extraction Of Weak Fault On Rolling Element Bearing

Rolling bearing is widely used in mechanical system. To realize the early fault diagnosis of rolling bearing, this has important significance to guarantee the safety of mechanical system. The papers according to the characteristics of incipient fault of rolling bearing signal characteristic is not obvious, introduces the advantages and disadvantages of the commonly used analysis methods of bearing fault, presents a deconvolution algorithm based on minimum entropy iteration, the algorithm can according to the signal characteristics of its own, through repeated iteration, to the input signal and output signal to noise ratio reaches a maximum for the termination condition, self-selection of the filter order adaptive, avoid the subjective factor influence of minimum entropy deconvolution algorithm by people’s shortcomings.In order to verify the feasibility of the algorithm of weak fault feature extraction and effectiveness, based on the correlation of rolling bearing vibration model on the deep analysis, separately from the signal to noise is relatively small and impact energy of smaller components of the two aspects of the simulation of fault signals, using iterative deconvolution method to achieve the extraction of weak fault feature. Finally, the proposed algorithm is applied to the experimental data, through the analysis of weak fault signal of rolling bearing, prove the effectiveness of the algorithm. The simulation and experimental analysis, and contrast the convolution method solution and the Hilbert envelope commonly used and minimum entropy results show that iterative deconvolution method can accurately extract the weak fault of rolling bearing characteristics, realize the early diagnosis of rolling bearing.