A Novel Neural Network Model of Hysteresis Nonlinearities

A novel and simple modeling method of hysteresis nonlinearities is proposed. Through analyzing the principle of the classical Preisach model, we find some characteristics and rules of motion point, i.e. trajectory of output to input, and believe that hysteresis curve, with analytic geometry method, can be constructed. The hysteresis curves from the constructed models, wonderfully match with a class of simulation hysteresis model, which consist of many backlash models. Though the hysteresis model is only a special class, when its output is used as one of input signals of neural networks, the neural networks model can approximate other classes of hysteresis curve. Three examples, including one simulation data set and two measured experimentation data sets, are implemented. The results indicate that the proposed method is successful and simple