Direct adaptive control of permanent magnet motors

This thesis examines the control issues associated with permanent magnet (PM) motors and provides speed control and torque control solutions using adaptive control techniques. Permanent magnet brushless de motor (PMBDCM) and the permanent magnet synchronous (PMSM) motor are studied. The PMBDCM has a winding pattern that produces a trapezoidal back emf waveform. The PMSM has a sinusodal winding pattern producing a sinusoidal back emf. The control objectives are to get precise speed control with reduced torque ripple and high motor and inverter efficiencies. Direct model reference adaptive control (DMRAC), self-tuning regulator (STR) and indirect model reference adaptive controller (IMRAC) are applied and these are compared with the proportional-integral (PI) controller.; The sinusoidal winding pattern of the PMSM allow us to use a two axis rotor reference frame for the control. The control parameters are the direct-axis current i_d and quadrature-axis current i_q. i_q is the main torque production component and i_d is flux weakening component. We have provided a new approach for controlling these two components.; The trapezoidal winding pattern of the PMBDCM does not allow us to use the rotor reference frame for the control purposes. The current magnitude level I_s is the control parameter. The generic inverter saturation problem of the PMBDCM which causes significant torque ripple is studied and we have developed a new approach to solve that problem. The new control strategy needs an extra control variable β which is the rising and decreasing angle for quasi-square motor phase currents.; To support credible control development, the work is focused on a commercially available 30 hp PMBDCM and PMSM. Finite element data form the basis of the motor model for use in the MATLAB Simulink and Saber circuit simulator. Experimental verification of the work is focused on the PMSM. Transient tests show that DMRAC is able update the controller gains to different operating points and system parameters without performing parameter estimation. Steady-state tests show that DMRAC is able to reject torque oscillations.