Research On Noise Reduction Method For Rolling Element Bearing Based On Vibration Signal

The rolling bear in the big rotation mechanical devices is the key but also easy damaged component. Once there is something wrong with the rolling bear, it will result in the unstable operation of the mechanical devices, or even more serious consequences. Fortunately, the vibration signal extracted from the fault components often demonstrates the feature of wave. The vibration signal of the rolling bear often includes a great deal of the operation state information of the bears, and thus it is one of the main basis for the further fault analysis of the bear. In this thesis, the noise signal reduction method for the simulation signal as well as the practical signal is investigated for the vibration signal of the rolling bear in the big rotation mechanical devices, and thus, it will be of great theoretical meanings and practical application values for the synthesis of the vibration signal, and also prevent safety accident.In this thesis, the main aim is to investigate the noise signal reduction method for the vibration signal fo the rolling bear in the big rotation mechanical devices, and then retrieval the required vibration signal. To deal with the unstable feature of the kind of vibration signal, the improved wavelet semi-soft shrinkage noise signal reduction is proposed. The signal retrieval is the key step to implement fault diagnosis for the mechanical devices, and thus it will be of great importance for the fault diagnosis of the vibration signal of the rolling bear.In the first place, the basic theory of the wavelet analysis is presented, followed by the basic definition and properties of the wavelet analysis, and the typical basic algorithms of the wavelet analysis is also introduced. Secondly, after presenting the traditional two noise signal reduction methods, including the hard shrinkage and soft shrinkage based on wavelet transform, and pointing out the advantages and disadvantages of them, an improved semi-soft shrinkage noise signal reduction method is proposed to deal with their defaults. The performance indices are also given to evaluate them. Thirdly, the three approaches are employed in the simulation as well as practical vibration signal of the rolling bear in the big rotation devices, the performance superiority of the proposed approach over the other two methods is validated by the performance indices computed and the frequency spectrum.Finally, the fault feature retrieval problem for the noise signal reduction of the vibration signal of the rolling bear is investigated, the Winger distribution and singularity value decomposition method are also introduced, the feature vectors are obtained for the vibration signal of the rolling bear in the four states, and thus the further noise signal reduction can be conducted for the rolling bear.