| Summary: | 碩士 === 逢甲大學 === 電機工程所 === 91 === This paper presents wavelet neural network with hybrid genetic algorithms (hybrid GAs) to design spindle motor speed controller for CD-ROM for solving nonlinear and uncertain problems those traditional controller can’t handle. Since the advantage of global search abilities of genetic algorithm and the ability of multiresolution analysis of wavelet transform and the learning ability of artificial neural network, such the hybrid GAs wavelet neural network has the power of learning arbitrary rapidly function. A hybrid GAs wavelet neural network speed controller of a three-phase, nine-slot and 12-pole spindle motor for CD-ROM drive has been designed. The experimental results are compared with close-loop fuzzy and traditional neural network speed controller. Experimental results demonstrate that the spindle motor speed controller using hybrid GAs wavelet neural network can reach its stable revolution speed more rapidly and has smaller steady-stead error, and reduce the arise time of motor speed.
Furthermore, this thesis presents Speed Monitoring system of Spindle Motor designed by wavelet transform. Since fault signal is temporary incident and this temporary distortion signal is usually not equal to zero in a tiny area, so it’s ineffectively for detecting temporary fault signal with traditional analytical method, such as Fourier transform etc. Owing to the wavelet transform has good natures both in time and frequency domain, and the characteristic of multiresolution analysis, so it’s an efficient method in detecting and localizing transient distortion signal. To achieve the request of real-time monitoring, we link up the program module built in DSP motion card with control program written in C language to build a Human Machine Interface for monitoring spindle motor speed at Simulink environment.
|
|---|
