Simulation Research for Giant Magnetostrictive Actuator Controller Using Model Reference Control Based on Neural Network

Giant Magnetostrictive Material (GMM) has inherent hysteretic nonlinearity, and its hysteretic performance changes with input frequency. Hence, it is difficult for a normal controller to control its actuator precisely. Due to this, a hysteretic compensation control strategy was proposed. Adopting neural network model reference, combine the dynamic model of Giant Magnetostrictive Actuator (GMA) as reference model, with BP neural network. Introducing error feed-back learning scheme - BP into controller and identifier, controller can identify GMA and identifier control it precisely. To accelerate the convergence of the trace error, train the neural network offline.