Indirect Adaptive Feedforward Control for Permanent Magnet Motors

Feedforward-feedback control can be used to achieve torque ripple minimization or error minimization between the reference and the output. Theoretically, if the inverse transfer function of the plant is known, it is possible to achieve zero error through feedforward control. In this paper, the references (acceleration, velocity and position) are designed so that they can be used as inputs to the feedforward model. A generic feedforwardfeedback control scheme for different back EMF shapes has been designed. It has been shown experimentally that the proposed feedforward scheme works well with the references. An error which is four times smaller is achieved using feedforward-feedback control compared to using feedback control only. Furthermore, the feedforward scheme is simply the inverse transfer function of the motor dynamics of a PM motor. It is simple and not computational intensive unlike most feedforward design. If there are any changes to J and B, it has been simulated successfully that the MRAS estimator is able to adapt accordingly, using the estimated J and B as inputs to the feedforward scheme. This allows a more robust feedforward control as a whole.