Hirokazu Ishihara

Simulation and experimental study on an air preessure control system by using a model prediction control.

In this parer, we deal With the process control system(SISO) and model prediction control(MPC) algorithm. The control results of MPC for a standard performance index do not behave as one expect. We study the MPC algorithm on a performance index with the control energy constraint. Furthermore, we propose an adaptive MPC algorithm with an estimation of the impulse response function by means of least squares. The steady state error is improved by the proposed method. The method is examined by simulation and experimental studies on the air pressure control plant.

Construction of PID Control System Based on Generalized Minimum Variance Control

PID control algorithms have been used extensively for several process control systems up to now. On the other hand, Clarke has proposed a generalized minimum variance adaptive control system with linear quadratic cost function as self-tuning control method for nonminimum-phase systems.In this paper, we consider the relationship between the PID control system and the generalized minimum variance adaptive control system. Then, it is shown that the PID control system is equivalent to a control strategy of minimizing a cost function in the generalized minimum variance control system. Furthermore, it is shown that PID control system can be regarded as a feedback control system with a pre-compensator. Next, we consider a design method of PID controller based on the generalized minimum variance control theory. Thus, PID gains are determined so as to construct a desired pre-compensator which suppresses an overshoot and shortens settling time or raise time. Finally, we show a numerical simulation result by the present method and apply the method to a pressure control system to show the effectiveness.

Multivariate discrete-time model reference adaptive control with input-smoothing property and its application

Multivariable discrete-time MRAC schemes that reduce the input vibration caused by small vibrating disturbances in plants are presented based on both indirect and direct methods. These schemes make it possible to insert low-pass filters in the transformation process from the disturbance to the input vibration without degrading the ability to track the reference model output sequence. Input smoothing performance is illustrated by application of the proposed schemes to a three-input three-output plant.