RECENT TECHNIQUES ON OBSERVER DESIGN FOR DISTURBANCE ESTIMATION AND REJECTION IN PERMANENT MAGNET SYNCHRONOUS MOTORS

Abstract

Permanent magnet synchronous machines (PMSMs) (either motor or generator) have attracted attention of research community comparing to other types of AC machines in the recent two decades. PMSMs are preferable than other AC machines in terms of large power-factor, broad speed of operation, compact proportions, and effective operation. Unfortunately, different sources of nonlinearities, model uncertainties, and external perturbations determine severity in a design of accurate speed control scheme for PMSMs. In the era of developing science and technologies, many advanced control solutions are proposed to control PMSMs. Although new solutions show their advantages comparing to traditional methods in terms of performance evaluation, practical realization of those algorithms could require expensive hardware with high computational capabilities. Furthermore, people in industry with less knowledge about the motor control may experience difficulties in using such advanced controllers on their own. Traditional PI/PID control schemes still work as a major control technique in modern industry, and in motor control as well. Numerous positive facts about the PI/PID schemes make such superiority of these control schemes. Firstly, the PI/PID can be implemented easily on most industrial software and hardware components. Secondly, while its scheme has clear mechanism of operation, most industrial processes could be controlled via the PI/PID scheme. These schemes are good in terms of small number of parameters to tune and tuning process itself could be very straightforward. Finally, implementation of the PI/PID controllers would require smaller time comparing to most proposed complex control solutions. It is studied that the traditional PI/PID controllers usually cannot deal with unpredictable disturbances, which in turn leads to degraded performance of an overall control system. Inspired by the advantages and widespread application of PI/PID control structure in industry, we propose a disturbance observer based composite control scheme which uses the PI-like controller for the feedback regulation and disturbance observer for estimation of lumped disturbances presented in a PMSM control system. Under this circumstance, this thesis work proposes three different control solutions for PMSM such as High-order disturbance observer-based composite control (HDOBCC), Disturbance rejection PI (DR-PI) control, and Hierarchical optimal disturbance observer-based control (HODOBC). Furthermore, to deeply understand the similarity and difference between the traditional disturbance observer-based control (DOBC) and active-disturbance rejection control (ADRC) schemes, this thesis also presents results of unification of these two control approaches in the speed control of a PMSM. The HDOBCC as the first method proposed in this thesis is designed to improve reference speed tracking performance of a PMSM under various operational conditions. A structure of the HDOBCC comprises a fuzzy-PI controller in a feedback stabilization part and novel high-order disturbance observer in a feedforward compensation part of the speed control system. The proposed controller is designed based on the research questions such as: firstly, although a fixed gain traditional PI controller is able to present satisfactory performance at some extent, still it does not guarantee such performance when sudden disturbances occur in a system; secondly, many disturbance observers designed for a PMSM in literature consider only a load torque as a disturbance, neglecting model uncertainties and parameter variations in design stage. Therefore, the HDOBCC is proposed such that it utilizes a fuzzy approach to determine parameters of the PI controller to overcome limitations of the fixed gain PI controller. Furthermore, the proposed scheme includes a high-order disturbance observer, which estimates not only the load torque, but also disturbances due to model uncertainties and parameter variations. Moreover, extended simulation and experimental studies are conducted to affirm performance of the HDOBCC under various form of the load torque. In addition to commonly tested step form of a load torque, severe forms of the load torque such as triangular form and sinusoidal form are tested with the proposed controller. Stability analysis of the closed-loop HDOBCC system is further provided. The next proposed method, DR-PI control, is designed by seeking answer for questions such as: firstly, although the traditional DOBC scheme applied for PMSM shows reasonable results in a PMSM control, its design can be limited to known actual parameters of the PMSM. In practice, actual parameters are usually not available, hence it could be hard to design the traditional DOBC in the absence of a plant information; secondly, for tuning a PI controller the traditional Ziegler-Nichols tuning approach still remains as one of the popular tuning approaches, however it does not give a rigorous explanation on selection of parameters during its design. Consequently, to answer these questions, the DR-PI control is designed for the PMSM speed control. The DR-PI control is designed such that it has a simple PI-like structure with intrinsic disturbance rejection mechanism determined by the parameters of a filtering element, desired plant model, and desired closed-loop system. Simulation and experimental validations are provided to validate the performance of the DR-PI. Furthermore, gain tuning mechanism and stability analysis of the closed-loop DR-PI-based speed control are also presented. The HODOBC scheme as a third proposed control scheme targets on the next research questions as: first, parameters of the traditional PI controller are mostly obtained by trial-and-error approach, which in turn may not guarantee satisfactory results; in a cascaded PMSM control, the outer speed loop performance highly depends on the performance of the inner current loop. The well-tuned speed control loop may degrade in performance, if the inner current loop is not tuned properly. To address these questions, we propose the HODOBC scheme, which consists of optimal PIlike controller in the feedback stabilization part and optimal extended-state observer (ESO) in the disturbance compensation part. The proposed HODOBC showed better performance when it is compared with other traditional controllers via experiments. Stability analysis is provided via the root locus approach. The study on unification of the DOBC and ADRC schemes has the following research question: the DOBC and ADRC are both used in estimation of total disturbance, but these two schemes are considered differently in literature. Hence, the study of both scheme is conducted to show the condition at which these two schemes show identical performance. The analysis of the traditional DOBC and ADRC schemes concludes that both scheme are equivalent in terms of performance characteristics if the dynamical delays of disturbance observers in each scheme are same. The results of analysis reveal that both scheme can be utilized to design a robust control system for PMSM, i.e. once the gains of disturbance observers can be calculated under the DOBC framework, further the disturbance rejection mechanism can be achieved via the ADRC framework. The results of PMSM control with the proposed control schemes have been tested on the Lucas-Nuelle DSP-based experimental setup.