Please use this identifier to cite or link to this item:
Title: Linear regression position estimation methods for switched reluctance motors
Authors: Chang, Yan Tai
Degree: Ph.D.
Issue Date: 2014
Abstract: In this research different sensorless methods that could reliably help to remove the need to use position sensors in switched reluctance motors(SRM) are developed. These methods are different to the conventional ones by providing unique rotor positions without relying on premeasured motor magnetic characteristics. The approach is systematic and mathematical. The relationship among motor phase inductance profiles is investigated and three types of functions to model their relationship are presented. Through regression analysis the function coefficients are found. Methods to linearize these functions for regression are explained. These coefficients are then used to calculate the rotor positions at standstill or flying restart. The estimation accuracies and efficiencies of the methods using the three functions are compared. These methods are collectively called the Linear Regression Position Estimation Methods(LRPEM). Two methods for position estimation on a running motor are also developed. A special event called Adjacent Phase Inductance Profiles Crossing(APC) when an SRM is running is used to turn the motor into a low resolution encoder. The speed and position values obtained from this encoder are fed to a Kalman filter where detailed rotor position values are produced. The low resolution encoder is used to commutate the phases with limited speed and torque capability, while the high resolution outputs from the Kalman filter are used for fine commutations. The two commutation methods are compared and the effects from mutual couplings, voltage drops and magnetic saturation are shown.
Subjects: Reluctance motors.
Hong Kong Polytechnic University -- Dissertations
Pages: xvi, 117 leaves : illustrations (some color) ; 30 cm
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of Nov 22, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.