The Research On Motor Bearing Fault Online Detection System

Rolling bearing is an important and vulnerable component of motor, a lot of motor’smalfunctions have related to bearings. The fact that about 40% faults have something to do withrolling bearings statistically indicates great influence that bearings have on motors. Bad design,installing inappropriately, bad using conditions, even sudden load could cause all kinds of problems,and change equipment’s statues for good. Therefore bearing fault analysis is always a studyinghotspot.Nowadays, main means of rolling bearing detection and fault analysis is vibration analysis.Usual procedure is sampling vibration signal and calculating characteristics of the signal, which tellbearings’status. The key is sampling and status reorganization. The dissertation is mainly aboutstudying bearing fault diagnosing technology, analyzing cause of malfunction, and constructingfaults’model. Diagnosing technology has improved rapidly these years, a lot of new methods havebeen developed, so based on current signal processing methods, some news mathematicstechnologies ,such as wavelet analysis , wavelet-packet analysis, empirical mode decomposition,have been used in the dissertation.First, structure model of bearing was studied, and several simple diagnosis methods wereintroduced, mainly about time-domain characteristics. Then a bearing fault detection system wasdesigned, software and hardware diagram included. Second, on the basis of time-domain diagnosistechnology, frequency-domain technology, like syntony demodulation technology, was used toanalyze a set of signal to verify its effectiveness. However, the syntony demodulation technologyalso has flaws, so in order to improve diagnosis’s accuracy, some news methods were used, likewavelet analysis, empirical mode decomposition, etc.The dissertation proved in theory and practices that using wavelet filter before syntonydemodulation technology could improve its flaws. Moreover, wavelet-packet band-pass filter couldmake syntony demodulation technology more effective. In the end, a new diagnosis technologycombined EMD and wavelet-denosing was invented, and its validness was provn in practice.