The Research To The Control Of BLDCM Based On ANN

Brushless DC Motor (BLDCM) not only retain the excellent dynamic and static speed regulation capability, it also hold the merit of simple structure, credible operation, little moment inert as well as easy control, it is a kind of relatively perfect operation component. But the traditional method has it's disadvantage of coarse mode, torque ripples (especially when the rotate speed below 100r/min) etc. So, it is suit to timing systems which has little request in speed standard only, to high precision servo system, the precision of traditional method is compared low much, this method must to be improved to adapt the need of servo system.SVPWM is used in this research to make the current in stator of the anti-electromotive force which is close to sine approach sine wave either, thus, toque ripples can be restrained. Whereas the localization in realization of SVPWM before (mostly is the trigonometric function table fixed), the ANN (Artificial Neural Network) is introduced to realize SVPWM, so the methods of Artificial Intelligence how can be used in power drive systems and the relation between SVPWM and torque ripples retrain are investigated deeply.In order to carry out sleazy, credible as well as simple SVPWM control in BLDCM, the six discrete Hallor-position signals to construct multi-polygonal, thus "angel-section compensation technique" can be used to obtain electric angle that SVPWM need.For these intentions, a whole hardware platform which is at the core of dsPIC30F6010A has been set up to realize the control to BLDCM, programmers have been written and debugged. The experiments proved that the idea of SVPWM based on ANN is feasible, torque ripples is reduced to about 60% of traditional control mode, flexibility and adaptability of the system are enhanced as well.