Micro Motor Fault Diagnosis Based On Wavelet Analysis And BP Neural Network

Micro-motor on-line fault diagnosis can feedback quality of the micro-motor timely, to facilitate the adjustment of the production process and production equipment debugging, which has a great significance for the development of micro-motor industry. Mechanical noise is the transmission result of mechanical vibration in the air or solid medium, frequency components of noise and vibration signal is basically the same, therefore, fault diagnosis through Sound signal is applied widely in which vibration signal is difficult to extract. To diagnosis micro motor fault in normal, stent deformation and iron-containing,a method analysis sound signals based on wavelet and BP neural network is presented in this paper.Industrially sound signal having a mixed complex features, the sound signal collected tend to have a lower signal-to-noise ratio. In this paper, used wavelet threshold de-noising method in micro-motor sound signal de-noising, to improve the fault signal-to-noise ratio is conducive to further fault feature extraction. Analysis method based on wavelet packet,in the case of no loss of signal energy of the signal is decomposed into a wavelet translation and scaling from the base family of functions. Obtained the exploded signal distribution in different frequency bands, conducive to a fault characteristic is separated from the complex signals. The energy on each Sub-frequency band, which are extracted by Wavelet packet decomposition and reconstruction algorithm are used to normalization process to establish energy and Fault mapping relationship. Finally, the fault characteristic parameters of the extracted sample input to the BP neural network algorithm based on adaptive learning rate training to establishment of the type of fault classifier, which can achieve the intelligent identification of micro-motor fault. Experimental results show that energy extracted by wavelet packet covers the micro-motor sound signal characteristics, while a well-designed BP neural network has a strong ability to identify faults, indicating that combined with wavelet and neural network diagnostic micro-motor fault through sound signal is effective. Based on the above research,the micro-motor fault diagnosis system has been design in LabVIEW platform,which system implements the micro-motor sound signal acquisition, display, storage, analysis,and micro-motor fault diagnosis.