Modeling of rate-dependent hysteresis using extreme learning machine based neural model

In this paper, a modified single hidden layer feedforward neural network (MSLFN) based model to describe the behavior of rate-dependent hysteresis inherent in piezoelectric actuators is proposed. In the proposed scheme, the improved SLFN model combining the weighted sum of simple backlash operators and the weighted sum of linear dynamic operators. According to the technique of the extreme learning machine, all the parameters of both backlash and linear dynamic operators are randomly assigned, while the output weights are determined by the least square (LS) algorithm. Then, the experimental results on a piezoceramic actuator are presented. It is shown that the improved model has obtained satisfactory approximation and generalization.