| Summary: | 碩士 === 中原大學 === 機械工程研究所 === 94 === This research discusses the adaptive digital filter applied to the zero across points detecting for the sensorless control in the brushless direct current motor by the counter electromotive force terminal voltage quantity measurement method.
When we detect the motor counter electromotive force to decide the rotor position, the external environment disturbance and the hardware uncertainties such as electric circuit noise and A/D-D/A transformed inaccuracy can affect the accuracy of the change phase signal. The situation is especially obvious when the motor is operating in the low rotational speed. In such a situation, the low voltages of the motor counter electromotive force combined with the oversized noises causes the zero across points detecting signals be disturbed and bring the erroneous judgments. Therefore, this research attempts to design an adaptive digital filter making the filter band-pass frequency change along with the motor rotational speed to reduce the effects of noise, and then the motor can be driven in a wider speed control scope with a better operating efficiency.
The adaptive digital filter mainly includes two parts: one is the digital filter structure; the other is the algorithm of the adjustment of the weighting parameters. The digital filter is constituted by the FIR filter. The adaptive deductive method is based on the least-mean-square method. We adjust the coefficients of the FIR filter by the calculated result of the least-mean-square method. Furthermore, we normalized the signals to increases the astringency of LMS method. Besides, the wavelet theory was the common used signal process method in the past few years; this research used the wavelet theory to analyze the counter electromotive force signals. The both results are compared to explain the potency of the digital filter.
The experimental results showed that the adaptive digital filter can filter the noise and disturbance and the phase lag can be effective improvement in the low rotational speed. The filter reduces the error decisions of the zero across points detecting signals; also the phase error was in an acceptable range of scope 30 degrees. The adaptive digital filter is not as good as the wavelet theory for the noise rejection. But both of difference was not clearly.
|
|---|
