Designing a Neural-Sliding Mode Controller for a Certain Class of Permanent-magnet Synchronous Motors for use in Unmanned Submarines

In this paper, a new design is proposed for a neural-sliding mode controller for a certain class of permanent-magnet synchronous motors to employ in unmanned submarines. Using presented specifications; first, the intended motor is designed and simulated in the Maxwell full-wave software based on the finite element method. Then, the nonlinear model is derived by the neural network using the data obtained from the simulation of the motor in the software. Since the model-based control methods are expensive and error-prone due to the high dependence on the modeling as well as the use of sensors, a sensorless controller is used, and with respect to the proposed nonlinear method, a sliding mode controller is designed for this particular motor. The proposed neural-sliding controller has obtained better results than the PID controller due to its robustness against system uncertainties and external disturbances, the overall closed-loop stability and convergence of error-trace to zero. The simulation results confirm the proper performance of the proposed model and controller.