Takashi Shigemasa

An application of state-space linearization to a power system stabilizer

A design procedure based on state-space linearization is presented for a power system stabilizer. The method involves approximations of complex nonlinear terms by quantitative evaluations, a state-space linearization which leaves the first order terms of the Taylor series from the nonlinear dynamics at equilibrium point, and optimal regulator design. The effectiveness of the procedure is shown by computer simulations.<<ETX>>

A model driven PID control system and its application to chemical processes

PID control systems are used as a basic control technology in contemporary industry. However, there are issues regarding control performances for some controlled processes, such as long dead-time processes and unstable processes. In order to improve the control performances of PID control systems, we proposed a new model driven PID control system, which is combined with the following control scheme; PD feedback, internal model control, set point filter and PV filter. In this paper, the model driven PID control system, its design method and application results for chemical processes are described.

A Kalman filter based robust model predictive control with constraints

Abstract In this paper, a new model predictive control method (MPC) derived from the modification of Generalized Predictive Control (GPC) is proposed. Firstly, a Kalman filter based predictor is introduced. Secondly, a time-dependent weighting factor is newly introduced into the MPCs cost function to improve the transient responses. The cost function is also extended by adding a new term related to the reference tracking error of the manipulation variables to handle redundant manipulation variables. Thirdly, a parameter tuning method considering robust stability of the control system is proposed. Consequently a constrained control type algorithm for a MIMO plant is formulated.

Identification and Control Experiments for a Large Space Structure

Abstract This paper presents results on studying the identification and attitude control problem for an LSS (large space structure). The proposed identification method is based on Biermans UD factorization method where the order of the identification model is determined by the frequency domain criterion, that is, the whiteness of the residual sequence. Two control methods, i.e. (i) I-PD control and (ii) optimal regulator are described. In particular, an optimal regulator was designed to locate all the closed-loop eigenvalues in a given region which may realize the desired dynamic behavior of the closed-loop system, so the control system has a large vibration suppression capability. The efficiency of these proposed identification and control method is shown by a ground-based test model.

A Design Method for Sampled-data Decoupled Control Systems with Multirate Sampling Periods

Abstract One of the authors has developed a practical design principle, called partial model matching. Applying the principle, a new design method for multi-input/multi output sampled data I-PD control systems is discussed. A special feature of this design method is that both the response shape and the sampling period can be selected independently for every control loop. Points in this method are given on how to realize the above feature. A digital I-PD control algorithm with multirate sampling periods is explained, after deriving design formulas. The effectiveness of this method is demonstrated by simulation tests.

A New PID tuning method by applying an iterative concept

Abstract In this paper, we propose a new PID controller design method based on the iterative identification and controller design scheme. At the identification part, we use a frequency weighted least square method for closed loop experimental data. The controller design is made two parts. (1) to design a frequency weighted LQG controller. (2) to design a PID controller by LQG controller reduction. In the reduction part, we use a frequency domain model matching method that is focused on the frequency response of the loop transfer function. The efficiency of this algorithm is shown through some simulation results.

Robust indirect adaptive control combined with frequency domain methods and its application to robot

The authors propose combining indirect adaptive control, which has wide applications, with frequency-domain methods and show the schemes application to a robot. In order to identify the wide-frequency response of a controlled system from the input-output data accurately, a multiple recursive identification method, combined with the decimation technique, is presented. The control parameters of a two-degrees-of-freedom PID control system, are determined by using the model matching method in the cutoff frequency region on the basis of the identified frequency response. Adaptive control based on frequency-domain methods can be widely applied to controlled systems, including not only minimum-phase systems but also nonminimum phase systems, even at the fast sampling rate. The superior features of the adaptive control scheme are summarized and its effectiveness is shown through experiments using an industrial robot.<<ETX>>

Model Predictive Control with Multi-Objective Cost Function considering Stabilization and Linear Cost Optimization

Abstract Almost all control problem are essentially multi-objective optimization problems which include such items as closed loop stabilization and economic cost optimization As a dynamic optimizer for the process operation, a new Model Predictive Control (MPC) algorithm based on the multi-objective optimization problem is formulated and analyzed. It uses an extended cost function including both conventional penalty terms for control errors or fluctuations, and newly introduced linear cost terms for output and/or input variables. The effectiveness of the proposed method, from the viewpoint of economic cost, is confirmed with some simulation studies, especially in the case of rapid change of the process operating conditions.

A Real-Time CAD Environments for Model Predictive Controllers

Abstract The model predictive control (MFC) technique has attracted attention recently as a practical process control method that realizes automatic plant operation. In this paper, a model predictive controller with CAD system developed by Toshiba Corporation named MIPCON (model identification and predictive controller) is introduced. Where, several functions which a real-time CAD system requires for process modelling and MPC design are discussed. Firstly, system concept is outlined and basic functions for modelling and controller tuning are discussed. Secondly, interface between MIPCON and conventional DCS and the man-machine interface are introduced. An application example to a distillation column model is also introduced.

IDENTIFICATION AND CONTROL EXPERIMENTS FOR A LARGE SPACE STRUCTURE

This paper presents results on studying the identification and attitude control problem for an LSS (large space structure). The proposed identification method is based on Biermans UD factorization method where the order of the identification model is determined by the frequency domain criterion, that is, the whiteness of the residual sequence. Two control methods, i.e. (i) I-PD control and (ii) optimal regulator are described. In particular, an optimal regulator was designed to locate all the closed-loop eigenvalues in a given region which may realize the desired dynamic behavior of the closed-loop system, so the control system has a large vibration suppression capability. The efficiency of these proposed identification and control method is shown by a ground-based test model.