Efficiency improvements in switched reluctance motor position and torque control using adaptive fuzzy systems

Switched and variable reluctance motors are well suited to use in direct-drive torque and position control of robotic actuators but suffer from nonlinear torque production characteristics. It has been demonstrated that adaptive fuzzy systems are capable of learning nonlinear current waveforms suitable for linearisation of the torque production characteristics in switched reluctance motors. This paper reports an investigation into the use of an extended heuristic method in order to produce solutions to the torque ripple minimisation problem that are particularly efficient with respect to copper losses. Simulation results based on experimentally measured data are presented demonstrating the influence of modified learning rate functions on the solutions learned by an adaptive fuzzy system and that these compare favourably with optimal solutions