Position Sensorless Detection of SRM Based on BP Neural Network

Theory of SRM position detection based on exciting pulse voltage to no conduction phase is analysised, and a combination with BP (back propagation) neural network model to predict SRM rotor position without sensorless is proposed. Training sample and test sample can be obtained by SRM detection system with position sensor, with response current as input, and the rotor position as output, a single input and a single output of BP neural network model is built. Mapping between response current and rotor position are obtained by training BP neural network, and the rotor position prediction is realized, test sample is used to verify the BP neural network model. Simulation results show that this approach improves the detection accuracy compared with the method by injecting exciting pulse voltage.