Realization Of Rolling Bearing Diagnosis System

Rolling bearing is an important part of mechanical and electrical equipments. According to statistics, the existing mechanical equipment failures,30% are caused by the rolling bearing failure, we can see their work directly affects the performance of machinery and equipment. Because the randomness of life of large rolling bearings, the timing is not suitable for traditional maintenance system. Rolling element bearing condition monitoring and diagnosis, changes in maintaining the condition maintenance or predictive maintenance to prevent performance degradation machinery and equipment capabilities, reduce or avoid the accident, has important practical significance.In this paper,we first introduced the vibration mechanism of rolling bearing and the function of characteristic and natural frequency to fault bearing, the indicators of time-domain parameters and the frequency domain parameter index; Study the method of based on wavelet transform and wavelet packet decomposition to extract fault features and constructed the initial feature set,because the mallat algorithm of discrete wavelet transform exist frequency alias, we put forward the improved single sideband reconstruction algorithm, provid the arithmetic software, Compare the performance of two algorithms.Explain the basic principles and calculation steps of principal component analysis and RBF Neural Networks, With examples of PCA-based feature extraction the necessity and effectiveness; Finally we built a framework for rolling bearing fault diagnosis system, using VC++and Matlab mixed programming language,developed rolling bearing fault diagnosis system software, using real data that gathered from Bearing Fault simulation sets the system functional testing, The results show that the system state recognition rate for the rolling bearing 90% or more, the better to complete the rolling bearing fault diagnosis tasks.