Research On Wavelet Neural Network In Electromotor Noise Fault Diagnosis Based On LabVIEW

With the development of modern science and technology, electromotor play a more and more important role in modern industrial plants. Electromotor will inevitably break down when it is running. In the course of operating, the electromotor not only will produce vibration, but also will send out strong noise, which contains abundant status information of the equipment, so we can use the noise signal to diagnose the fault. As the requisition of noise is rigorous in recent years, the noise of electromotor has been an important factor to influence on market benefit. Abundant research on the principle of noise and measure of noise elimination has been carried on, but consumers still reflect that the domestic electromotor also have the problem which makes the people feel the noise is too high, and they are often unwilling to choose domestic electromotor for this reason. So the electromotor producers have to promote the competence of fault diagnosis, look for a new method in this field, in order to improve the quality of the domestic electromotor and economic benefits.The paper discusses about how to research a noise fault diagnosis system on Wavelet Neural Network (WNN) based on Virtual Instrument (VI). As the fault signal is non-stationary, transient one, the traditional signal analysis methods, such as FFT are not so efficient and useful for the fault signal detection. However, Wavelet Analysis has the excellent time-frequency local performance, it can detect the different frequency components of the fault signals by its adjustable time-frequency window. In view of the superiority of Wavelet Transform to non-stationary signal, we introduce the principle of wavelet to eliminate signal noise and extract its features. Firstly, we use Mallat algorithm based on principle of multi-resolution to eliminate noise, and then decompose and reform the noise signal, deal with the coefficient of high-frequency, extracting the characteristic vectors as the input signal of the neural network. Using the...