Abstract
This paper proposes a method of identification for nonlinear system with saturation and hysteresis characteristics. Gears and wires, etc. are widely used for mechatronics equipment, mechanical systems, robots, etc. as transmission and converter of the power. In such transmission and converter of power, when especially, the direction of the force changes, the hysteretic behavior called the hysteresis characteristic is generated. The hysteresis characteristic is nonlinear two-valued function, and it is expressed by using the recurrent neural network. And, actuators are also show the nonlinear characteristic such as the saturation. They are modeled as the dynamic discrete-time system subject to gain saturation. The parameters of these models are identified based on input-output data of overall system which is combined dynamical system with the saturation and hysteresis characteristics. In the evaluation function for parameter estimation, the square error of system amd model is used, and the minimization of the function is carried out by the nonlinear optimization techniques.
