Wavelet Entropy And Support Vector Machine Are Used In Fault Diagnosis Method' Research And Application

With the rapid development of modern technology, engineering systems'complexity continue to increase, system'reliability and security becoming the key factors to ensure economic and social benefits ,so fault diagnosis technologies have been widely applied in various fields.At present, the more applicated fault diagnosis methods have the methods such as the Fourier transform and wavelet transform that based on transform methods,artificial intelligence methods,Porny analysis and the method who based on mathematical morphology . According to these methods'the principles and characteristics ,they have their own advantages in different areas .Based on the existing method , this paper is study a new fault diagnosis method which is based on the improved wavelet energy . I had systematic study the continuous wavelet transform,discrete wavelet transform,multi-resolution analysis,wavelet entropy and some other theories that base on wavelet transform theory, combined with the wavelet'noise separation technology,time - frequency localization characteristics technology ,treat the test signals with signal feature extraction technolopy and wavelet multi-resolution decomposition and reconstruction technolopy , from the time - frequency domain-binding analysis view, ues the sliding window algorithm, choose to use high-frequency band of wavelet coefficients feature extraction, study the improved wavelet energy, and using this method get the corresponding entropy value ,use these value as the amount of disturbance characteristics. To support vector machine as classifier, the fault identification .Use common faults in real life, for example, to simulate , and analyse the simulation results. The simulation results show that the method improves the accuracy of fault diagnosis.The method is an effective method.