Vladan Petrovic

Modeling and Minimization of Torque Ripple in Permanent Magnet Synchronous Motors

May 26, 1998
1:30 PM
106 Egan

Abstract

Permanent magnet synchronous motors (PMSMs) are rapidly gaining popularity in high-performance, variable frequency drive applications such as machine tools, and direct drive robotics. In such applications, the production of smooth torque is crucial. This thesis addresses the modeling of torque ripple and the design of several controllers for its minimization. Since standard PMSM models neglect the torque ripple, the phenomena responsible for torque ripple generation (such as non-sinusoidal distribution of stator windings and rotor magnet, and existence of stator slots) were analyzed first and included in PMSM model. After the field distributions in practical machine were considered and the model in the physical (abc) frame was derived, Blondel-Park transformation was used to convert it to a more compact form in the rotating (dq) frame. The accuracy of the resulting model was evaluated using numerical simulations with particular PMSM parameters, and comparing the results with experimental data measured for the same motor.

The verified model was used in the design of various controllers for torque ripple minimization. First, two classical linear controllers were designed as a benchmark for the following more complex algorithms. Since some model parameters may change with shifting operating conditions, two adaptive control algorithms were developed next, using the Lyapunov and passivity approach. Stability and parameter convergence were mathematically established, and controllers were verified by both simulations and hardware experiments. The experimental results show considerable reduction in main torque ripple components, but also point out some algorithm sensitivities to hardware imperfections.

Thesis Committee:
Prof. Aleksandar M. Stankovic (advisor)
Prof. Bahram Shafai
Prof. Gilead Tadmor