FPGA-Based High Performance Sensorless Control for PMSM Drives

This work focuses on the development of high performance sensorless controllers for permanent magnet synchronous machine (PMSM) drives. To realize this goal, two aspects are investigated. One is the improvement of control algorithms. The other is the exploitation of capabilities of field programmable gate array (FPGAs) which are the most promising candidates for the future drive control platform, to advance implementation performance.

The main contributions of this work are as follows:

A linearly compensated sensorless flux observer based on back electromotive force (EMF) is proposed. This flux observer exhibits excellent dynamic and steady-state performance, which is robust to parameter variation. It is effective with 0~3 times stator inductance variation at the speed above 0.03 pu (per unit). At the speed of 0.03 pu, the flux observer works well with 0~2 times resistance variation with and without load. As the speed increasing, the sensitivity of the flux observer to stator resistance variation decreases.

Using Delta-Sigma modulator, based on FPGA current oversampling and signal processing techniques, high resolution of digital current signals is obtained. Therefore, a lower high frequency (HF) current response is sufficient to implement HF voltage injection sensorless control methods. This can also result in lower acoustic noises, less losses and disturbances of the system. Using the designed FPGA controller, the injection could be reduced to one third in the standard drive system.

For the extended Kalman filter (EKF) based sensorless observer, an interesting result has been found. It demonstrates that different frame models have notably various influences on the implementation, especially for the implementation of FPGA. For instance, the well-known simplified rotating frame EKF observer cannot work properly. This problem can be found immediately in the implementation of FPGA based fixed-point data type. Thus, an approach using the hybrid rotating frame model is presented in this work to solve the problem.

A continuous set nonlinear model predictive control (CS-NMPC) for PMSM drives is proposed. It is easy to include nonlinearities and constraints of the control system. Therefore, the control system has not only high dynamic performance in the transient state, but also excellent performance in the steady state. Moreover, it is very easy for the control system to combine the sensorless control methods in realizing sensorless control for PMSMs. A model based design (MBD) method for fully FPGA based drive system controller is presented. Using Matlab/Simulink and HDL Coder, machine control and sensorless control methods are designed and implemented in a single FPGA chip. Based on fast parallel processing of FPGAs, sensorless control algorithms are implemented without disturbance of the execution of main control algorithms. High performance sensorless control is obtained due to high sampling frequency and low excitation time. It is reasonable to believe that FPGAs will become the core of drive control platforms in the near future, and the MBD method for FPGA based controller design presented in this work will receive more attention.