Zhang Zhongjun

Model predictive control for a class of nonlinear systems

In this paper a two-level model predictive control algorithm for affine nonlinear systems is presented. A static nonlinear state feedback is at first designed to obtain an I/O linearized closed-loop system. A model predictive control algorithm is then adopted for the closed-loop system to achieve good dynamics and robustness. The effectiveness of the algorithm is also verified by the simulation example of an inverted pendulum on a cart.<<ETX>>

A Computer Control System in Steelmaking with Different Control Strategies and Expert Knowledge

Abstract The paper presents a computer optimizing and supervisory control system with a knowledge base for the steelmaking process of an electric are furnace(EAF). The design of the control strategies, hardware and software are included. Three mathematical models are established for the different stages of the stoelmaking and three strategies are employed respectively. An expert control strategy is also used for practical use in the steelmaking shop. The system provides a user-friendly interface for operator. It is tested in practice and the results are very satisfied.

Predictive Control of Nonlinear Systems based on Extended Linearization

Predictive control for nonlinear systems of ten deals with nonlinear programming and is not suitable for real-time control. In this paper a multi-layer design procedure of the predictive control for nonlinear systems is presented. At first the nonlinear state feedback of the extended linearization is used to obtain the approximate incremental linearized model of the system, and then the predictive control in state space description is adopted to achieve fine tracking performance and robustness. The new design method is convenient for calculation and can meet the requirements of the real-time control. A simulation example is also given to verify its good performances.

A New On-line Structure Selection Algorithm for MIMO State Space Models

Abstract MIMO linear system identification is usually developed in a canonical form It has been pointed out that sometimes there exist some numerical problem concerning system identification because of its uniqueness. The overlapping structure model is suggested in 1974, since then, several structure selection algorithms have been developed. A new on-line structure selection algorithm for MIMO linear systems is described in this paper. It uses the average diagonal elements of the inverse of the information latrix (EVN) as measure of the conditoning of parametrization. According to the concept of the relative correlativity, the most independent vectors are seccessively selected from the Hankel matrix of the system with a modified Gram-Schmidt orthogonalization method. The structure of the better conditioning of the parametrization is thus obtained. The relation of overlapping parametrization between state space and the I/O difference equation models is derived, so that the realization of overlapping parametrization of state space form can be obtained by simple elementary operations. Combining any recursive algorithm for parameter estimation (e.g. RLS, RIV, RGLS etc.) with the algorithm of structure selection in this paper, on-line algorithm for overlapping state space model is readily developed. Computer simulation results show the applicability of the above algorithm.

The Synthesis of Decentralized Control Systems with the Most Economical Information Structures

Abstract In this paper, the existence of the solution for the Most Economical Information Structure (MEIS) problem is discussed. A criterion for determining the fixed modes as well as their orders, applicable to arbitrarily constrained feedback structures, is then given. This criterion provides the possibility to evaluate the contributions of all single feedback elements and their combinations to eliminating the open — loop fixed modes and thus furnishes the necessary information for synthesizing the MEIS. Based on the results, a recursive algorithm for synthesis of MEIS is presented and the design procedure of decentralized control systems with the MEIS is then proposed. Finally, an explanatory example is involved.

A New Criterion and a New Algorithm for Fixed Modes in Decentralized Control

Abstract Based on Matrix Fraction Description (MFD) and decomposition of the feedback structure into fundamental substructures, a new criterion for testing fixed modes as well as their orders w.r.t. arbitrary constrained feedback structure is presented, which leads straightly to a new algorithm for evaluating fixed polynomials and fixed modes without solving high order characteristic equation. Some new concepts are introduced to illustrate the successive procedure of the open-loop modes changing from fixed to variable, which provide detailed structural information about the occurence of fixed modes.

A Method for Modelling with Missing Observations

Abstract This paper proposes a new method for establishing discrete models with missing observations. The authors firstly develop a formula for calculating the unbiased estimator of the variance of noise series {e t }; then the model parameters are estimated by Monto-Sarlo statistic method. This proposed method is suitable not only for establishing AR models in time series analysis, but also for establishing ARMA models. For input-output descret systems with missing observations in input and output data, this method is also appplicable for modelling.

On Line Identification of MIMO Linear Discrete Stochastic Systems

Abstract In this paper the structure Identification and parameter estimation of MIMO linear discrete stochastic systems are discussed. Using Luenbergers observable canonical form with steady-state Kalman filtering representation, an on-line identification method developed. The structural identification of the systems is determined by residual error method by developing a recursive algorithm. Parameter estimation is made by recursive extended instrumental variable method which gives an asymptotic unbiased estimate. A recuesive algorithm for the residual-square-sum is thus obtained. Computer simulation results indicate that the proposed methods are sufficiently well to be used for adaptive control purpose.

A Recursive Algorithm by using Eigenvector Method for Identifying Multivariable Linear Time-Invariant Systems

This paper discusses the identification of multivariable linear time-invariant systems, under the condition that colored noise disturbance exists both in the input and output observation data simultanuously. A recursive algorithm is developed by using eigenvector method for identifying multivariable systems, and also the problem of convergence is discussed. The result of simulation on digital computer shows that the convergence of the algorithm is assured, even under the condition that the noise to signal ratio is up to 33% in the output and up to 20% in the input, satisfactory result can still be obtained by using the recursive algorithm given in this paper.