HIGH PERFORMANCE CONTROL OF PERMANENT MAGNET SYNCHRONOUS MOTORS

Abstract

High performance speed control of permanent magnet synchronous motors (PMSM) is considered in this research work. Since the model of PMSMs is nonlinear, controllers that are based on linearizing nonlinear models about an operating point, may not give an acceptable dynamic performance and robustness for the entire operating range of the PMSM servo drive system. Thus, there exists a strong motivation to apply nonlinear control techniques for high performance PMSM speed control. Several nonlinear control methods have been investigated and reported in the literature. This research work presents a nonlinear control technique based on differential geometry theory called Input-Output Feedback Linearization (IOFBL) for speed regulation of PMSMs. Although the theory of IOFBL is well developed, its practical application to PMSM speed control has been very limited. This research work attempts to contribute to the excellent work in PMSM control available in literature, by accurately modelling the PMSM simulating the IOFBL controller with this PMSM model, and verifying the model by obtaining experimental results and comparing it to the simulations. From these results, some conclusions can be drawn on the performance of the IOFBL controller. To understand the advantages of the IOFBL controller for speed control, it is also compared with a 3-loop proportional-integral (Pl) controller which is based on Jacobian linearization. The IOFBL and PI control algorithms are implemented on an experimental set­up that uses a digital signal processor (DSP) (TMS320C50) based digital controller. The experimental set-up designed in this research work consists of, a sine-triangle PWM based power converter with its associated gate drive and overcurrent protection circuitry, and an interface circuit between the DSP and sensor/power circuit. Vector control technique is used in this system and the 3-phase a-b-c variables are transformed into d-q variables using Park's transformation implemented by a software algorithm in the DSP. The controller algorithms are then implemented on these d-q variables by another software module within the DSP before being transformed back to the a-b-c output variables. The outputs of the DSP are then converted to analog sinewave reference waveforms that are used for PWM comparison. The performance of each controller is evaluated by studying the transient speed response to step changes in the reference speed and load torque. The responses of the d-q axis currents are also shown for comparison with the corresponding simulation results. The performance of the controllers for speeds above the base value is shown by operating the motor in the field-weakening region. The sensitivity of each controller to changes in motor parameters like back-emf constant, is also investigated. Performance figures such as percentage overshoot/undershoot, rise time/fall time and settling time are used for benchmarking the controllers. The results are analysed qualitatively, and quantitative values for different performance figures are listed. The experimental results obtained in this research work leads to the conclusion that the nonlinear IOFBL controller shows an improved performance compared to the PI controller for speed control of PMSMs. However, the IOFBL method is not inherently robust against parameter variations and modelling inaccuracies. As a result, additional algorithms may need to be developed, to improve the robustness of IOFBL controllers.