Yasuhide Kobayashi

Identification of nonlinear systems with hysteresis characteristics

This paper proposes a method of identification for nonlinear system with saturation and hysteresis characteristics. Gears and wires have been widely used for mechatronics equipment, mechanical systems and robots 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. This characteristic is a nonlinear two-valued function, and it is expressed by using the recurrent neural network. Actuators 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 the input-output data of the overall system which is a combined dynamical system with saturation and hysteresis characteristics. In the evaluation function for parameter estimation, the square error of the system and model is used, and the minimization of the function is carried out by nonlinear optimization techniques.

Derivation method of optimal solution on output feedback control

This paper presents a modified version of the output feedback control problem. The conventional output feedback control methods are semi-optimization methods that need to calculate repeatedly when derive an optimal output feedback gain because the Riccati equation includes a non-linear part. We considered a solvability condition that output feedback control has a unique solution. As a result, we obtained the stationary condition depending on sensors and including the same form of the Riccati equation as the well-known optimal regulator.

Robust Critical Control for Servo Systems using Disturbance Observers

This paper develops a computationally efficient method to design a controller for servo systems which require that tracking errors and control inputs are strictly maintained within pre-specified constant bounds. To ensure these bounds in the presence of unknown disturbances and parameter variations of a plant, the disturbance observer is introduced. In the framework of the principle of matching, a controller is designed as a 2-degree-of-freedom controller including the disturbance observer under the condition that reference signals and disturbances are restricted in rate of change and magnitude, respectively. Through a case study of servo controller design for the actively-controlled bed for ambulances, it is shown that the 2-degree-of-freedom controller can compensate wider parameter variations compared to a 1-degree-of-freedom controller and that the proposed method is computationally efficient

Identification of Linear System by Neural Networks Under Unknown Noise Density

Abstract The paper considers a problem of the modelling and the parameter estimation for a linear dynamical system, where the density of an observation noise is not available. The system parameters are estimated by a least squares method for candidate models, for the probability density and statistical values are not available here. The system structure is estimated by using a neural networks. The learning rules are used batchwise for candidate models with typical types of distributions. The connection coefficients of the neural network are optimized by the groups consisted of known structures. In the real case, since the system structure is unknown, the cases that the true system structure is not contained in the group for learning data are considerd.