Нейроестиматор для бездавачевої системи керування вентильним реактивним двигуном

摘要

To create a sensorless control system of switched reluctance motors with a more simple structure of the neural network than existing ones, a feedforward artificial neural network, that uses only phase currents as the input signals, was synthesized. It is proposed to apply the smoothing of the neural network’s output data to simplify its structure. Neuro-controller was synthesized to form a flat speed-torque characteristics of switched reluctance motors. By computer simulation the work of the sensorless control system using created artificial neural network was researched.