Abstract
研究就小型軸向氣隙型無刷直流馬達，因霍耳元件與相線圈偏移造成的速不平穩與轉矩漣波源做鑑別，同時對於轉速不平穩之現象設計控制器降低馬達轉速變動率。賑膍s第一部份將分別對於霍耳元件與相線圈因組裝上的問題所引起的轉x漣波源提出鑑別之方法；由於霍耳元件與相線圈偏移引起的轉矩漣波之峖’酗@固定週期與規則可循，因此可利用觀察量測到之轉矩漣波，判斷O屬於霍耳元件或相線圈之偏移所造成，如此可改善生產線之組裝精度。賑膍s的第二部份將就軸向氣隙型無刷直流馬達應用在低轉速時之轉速不倥酊滌暋D，提出一個控制器架構，以期能降低馬達之轉速變動率；因馬F本身的非線性行為，本研究應用類神經網路之技術，架構一結合類神經蘢糮e饋控制器與傳統比例迴授控制器之適應性控制器，以期能降低馬達b低轉速時的轉速變動率。This thesis focuses on the torque ripple analysis andvelocityontrol for a specific class of BLDC motors, i.e. axial-fieldLDC motors. For those kinds of motors, there are twosourcesf torque ripples which are neglected by previousliterature,.e. bias of Hall sensors and bias of armaturewindings.imulation of the torque production in such motorsindicatehat both of the bias could induce significant torqueripples.ecause the patterns of these torque ripple have fixedrule andixed cycle to follow. The first part of this thesisproposed aethods based on patterns of the ripples which arehere to do thedentification of such ripple sources. The methodscould supplyssembly processes of the motors with correctiveinformation.ext part of this thesis, a control structureincorporated neuraletworks is addressed, which is basicallyalearningcheme and has several advantages worthy of notice.Experment re presented, which verify the feasibility of thecontroller. his controller could effectively suppress thespeed fluctuationhether or not there is bias of sensors orwindings.