Parameters Identification on the Improved Hysteretic Preisach Model Based on the RBF Neural Network

摘要

To solve the problem of the hysteretic character of the piezoelectric material in real application, some measurements such as the initial weighted factors of the hysteretic units being calculated based on the theory of Preisach and the FORCs test data, the RBF neural network being introduced to calculate more weighted factors of the hysteretic units, the quadratic polynomial interpolation method being applied to improve the output of the hysteretic model and the Simulink tool being used to establish the simulation model were used to establish an improved hysteretic Preisach model. The parameters identification results of improved hysteretic Preisach model show that the hybrid RBF method is of accuracy according to the real actuator, the mean absolute error is about 1.0 μm and the maximum error is around 2.0 μn. Thus the hysteretic phenomenon is eliminated through the hybrid RBF method, not only the computational time is reduced, but also the accuracy of model is high. So the introduced method of RBF is an effective method for parameters identification of hysteretic Preisach model.