NO FULL TEXT AVAILABLE. Access is restricted indefinitely. ----- With the rapid development of power electronics and microprocessor control techniques,
various advanced drive schemes, such as brushless DC (BLDC), vector control (VC), and
direct torque control (DTC), for permanent magnet synchronous motors (PMSMs) have
been developed and extensively employed in drive systems of both industrial equipment ·
and domestic appliances because of the high power density, high torque to current ratio,
and high power factor. In these drive schemes, the position of the rotor magnetic flux is
required in order to' apply the stator flux vector for maximum torque. The conventional
drive systems use mechanical position sensors to acquire the rotor position. In recent years,
the sensorless control of PMSMs has attracted a great attention in order to eliminate the
disadvantages of using mechanical position sensors, such as the increase of hardware
complexity, cost and reduction of reliability.
Among these advanced drive schemes, the recently developed DTC scheme features in
simple control algorithm and excellent dynamic performance. The DTC algorithm detects
the position and speed of the rotor magnetic flux from the electromotive force (emf). This
method works well at medium and high speeds. When the rotor rotates at a low speed or is
stationary, this method fails because of the insufficient or zero emf signal. In order to start
the DTC drive system for PMSMs under load, it is essential to know the accurate initial
Most of the approaches to detect the initial rotor position make use of the structural and/or
magnetic saturation saliencies, and therefore, they are sufficiently effective for a motor with
large structural saliency, for example, the interior PMSM (IPMSM). However, in a surface
mounted PMSM (SPMSM), to detect the initial rotor positions is a great challenge due to
the very small structural saliency of the rotor.
This project aims to develop a high performance sensorless variable speed drive for
SPMSMs. This thesis includes two major parts: (I) modelling of the PMSMs with
structural and saturation saliencies, and identification of the nonlinear parameters taking
into account the saliencies, and (2) development of schemes to estimate the initial rotor
position without additional hardware for off the shelf SPMSMs with inherently small
saliencies compared with those of IPMSMs.
Due to the lack of a PMSM model that incorporates both structural and saturation saliencies,
the researchers in this field have been using the experimental trial and error method in
developing techniques for detecting the initial rotor position, which is inefficient. In this
thesis, a novel nonlinear model of PMSM incorporating both structural and saturation
saliencies has been developed and employed in the development of new initial rotor
position estimation techniques by numerical trial and errors. The method for measuring the
model parameters, e.g. the nonlinear inductances, is presented. The validity of the model is
confirmed by comparing the dynamic simulation results based on the proposed model with
those based on the model in the Matlab/Simulink library.
With the proposed model and the measured parameters, several initial rotor position
estimation schemes especially for the SPMSM have been numerically simulated. In order to
confirm the effectiveness of these schemes, experiments have been carried out to estimate
the initial rotor position by applying various types of voltage pulses to a test SPMSM. It is
found that the peak stator currents corresponding to the high voltage pulses are strong
enough to produce reasonably large magnetic saturation saliency for identifying the rotor
position and polarity of the SPMSM at standstill. A simple numerical algorithm for
implementing the high voltage pulse method is proposed. By analysing the factors related
to the current response of the voltage pulses, the method with the high voltage pulses to
amplify the saturation saliencies is refined. Based on the refined corresponding peak
currents, another initial rotor position estimation algorithm is investigated. The validity and
accuracy of both the schemes are confirmed by experimental results.
Also based on the proposed model, the performance of a conventional DTC scheme is
analysed by simulation and experiment. Finally, the presented initial rotor position
identification strategy has been implemented in a sensorless DTC drive for the SPMSM.
Experiments are conducted to confirm the improved performance of the sensorless PMSM
drive and the accurate estimation of initial rotor position.