Embedded Controller for Reduction of Torque Oscillations in a PMSM | Consortium for Embedded Systems | SIU

Abtstract: The proposed project will be an extension of a series of previously funded by the center projects addressing the control of a permanent magnet synchronous motor (PMSM). The ultimate goal is to develop a multiple function controller that can be incorporated into an integrated system-on-chip (SoC), thereby reducing hardware size and complexity. In the previous projects, two controller functions were developed for the PMSM: a speed regulator and an adaptive scheme to cancel stator current harmonics. In this project, it is proposed to investigate the torque harmonics generated at low frequencies as a result of a non-sinusoidal flux distribution in the motor. Subsequently, it is proposed to design a control loop to reduce the torque harmonics. The additional function will be incorporated into an SoC along with the previously developed subsystems of the controller.

Problem: PMSMs generate harmonics due to the imperfect distribution of magnetic flux. Harmonic generation is particularly pronounced in trapezoidal flux motors. As a result, harmonic stator currents will develop. The interaction between these harmonic currents and the harmonic emf produces torque oscillations at related frequencies. The main challenge in designing a suitable algorithm for reducing the torque harmonics is the dependence of the fundamental of the harmonic frequencies on the speed of the motor. Therefore, a conventional fixed frequency filter will not be effective. The approach must be based on adaptive filtering techniques. However, the complexity of the technique could impede implementation on an SoC, [1].
The proposed work will investigate adaptive methods for reducing the torque harmonics produced by a non-sinusoidal flux distribution in PMSM. The final algorithm will be optimized for implementation in a SoC in order to reduce hardware complexity.

Rationale / Approach: The main approach will consist in developing a suitable algorithm to counteract the torque oscillations due to the non-sinusoidal emf of the motor. The algorithm will be based on motor emf and torque estimation on-line. The proposed project will attempt to estimate the motor torque either based on direct dc current measurements or based on the stator current measurements. The estimated values of the back emf and torque will be used as feedback to counter modulate the stator voltage (through the motor inverter output).

Novelty: The main novelty of the project is in the method to estimate the motor torque without requiring a mechanical sensor. Another novelty is the incorporation of a complete motor controller into an SoC.
Potential Member Company Benefits: The primary benefit to the industry is the potential to reduce torque oscillations and to reduce the size of needed hardware.

Deliverables for the proposed year: The main deliverables include:
1. the code for an SoC implementation;
2. a MATLAB models of the motor and the controller for using in studies and assessing effectiveness.