Neural network application to an optimal control of a variable reluctance motor

The authors describe the problem of designing the real-time optimal efficiency control of a variable reluctance motor operating at the desired torque speed. An application of a neural network is proposed to perform an optimization of the motor excitation such that the drive efficiency is maximized subject to the torque speed requirements. The most interesting feature of the proposed neuromorphic controller is the possibility of adapting the computation of the optimal control in real time for the drive system. The neural network has shown its ability to identify the motor behavior, with its highly nonlinear characteristics, as well as to control its operation at optimal efficiency