Model-Based Position Estimation for Synchronous Reluctance Motor Drives

Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Toni Tuovinen Name of the doctoral dissertation Model-Based Position Estimation for Synchronous Reluctance Motor Drives Publisher School of Electrical Engineering Unit Department of Electrical Engineering and Automation Series Aalto University publication series DOCTORAL DISSERTATIONS 86/2014 Field of research Industrial Electronics and Electric Drives Manuscript submitted 27 March 2014 Date of the defence 20 August 2014 Permission to publish granted (date) 13 May 2014 Language English Monograph Article dissertation (summary + original articles) Abstract This thesis deals with model-based position-estimation methods for synchronous reluctance motor drives. The position estimation methods should be robust against modeling uncertainties, and reliable position information should be obtained at all speeds, including standstill. Since the considered position-estimation methods are model based, methods to obtain the model parameters under normal operation of the drive are proposed. Three observer structures are studied: a first-order observer, a reduced-order observer and a speed-adaptive full-order observer. The stability of the observers is studied via small-signal linearization under erroneous model parameters. Based on the analysis, robust gain selections are proposed with maximal tolerance for model parameter uncertainties. For improved performance, the observers are augmented with parameter adaptation laws. The first-order observer and the reduced-order observer are augmented with resistance adaptation laws at low speeds, using information from the back-electromotive force. The speed-adaptive full-order observer is augmented with inductance adaptation law at high speed, using information from the back-electromotive force. At low speeds, the speed-adaptive fullorder observer is augmented with resistance adaptation law, using information from highfrequency signal injection. The stability of the augmented observers are studied via smallsignal linearization under erroneous model parameters. Based on the analysis, stabilizing gain selections are proposed. The observers are experimentally evaluated using a 6.7-kW synchronous reluctance motor drive. The first-order observer seems unsuitable for these. The behaviour of the reduced-order observer is closely related to the behaviour of the speed-adaptive full-order observer, but the speed-adaptive full-order observer can be made less sensitive to parameter variations and measurement noise due to the additional degrees of freedom. When the speed-adaptive fullorder observer is augmented with parameter-adaptation laws, small position-estimation error is obtained at all speeds, including standstill.This thesis deals with model-based position-estimation methods for synchronous reluctance motor drives. The position estimation methods should be robust against modeling uncertainties, and reliable position information should be obtained at all speeds, including standstill. Since the considered position-estimation methods are model based, methods to obtain the model parameters under normal operation of the drive are proposed. Three observer structures are studied: a first-order observer, a reduced-order observer and a speed-adaptive full-order observer. The stability of the observers is studied via small-signal linearization under erroneous model parameters. Based on the analysis, robust gain selections are proposed with maximal tolerance for model parameter uncertainties. For improved performance, the observers are augmented with parameter adaptation laws. The first-order observer and the reduced-order observer are augmented with resistance adaptation laws at low speeds, using information from the back-electromotive force. The speed-adaptive full-order observer is augmented with inductance adaptation law at high speed, using information from the back-electromotive force. At low speeds, the speed-adaptive fullorder observer is augmented with resistance adaptation law, using information from highfrequency signal injection. The stability of the augmented observers are studied via smallsignal linearization under erroneous model parameters. Based on the analysis, stabilizing gain selections are proposed. The observers are experimentally evaluated using a 6.7-kW synchronous reluctance motor drive. The first-order observer seems unsuitable for these. The behaviour of the reduced-order observer is closely related to the behaviour of the speed-adaptive full-order observer, but the speed-adaptive full-order observer can be made less sensitive to parameter variations and measurement noise due to the additional degrees of freedom. When the speed-adaptive fullorder observer is augmented with parameter-adaptation laws, small position-estimation error is obtained at all speeds, including standstill.