J. Lieslehto

A Comparative Study of Some Multivariable PI Controller Tuning Methods

Abstract In this paper four different multivariable PI controller tuning methods are compared. Multivariable PI controller is designed using all four methods for a two input two ouput state state space model. The resulted designs are then compared using frequency domain performance and robustness criteria and time domain simulations

PID controller tuning using evolutionary programming

The evolutionary programming algorithm for PID controller tuning presented offers several advantages. One can use a high-order process model in the tuning, and the errors resulting from model reduction are avoided. It is possible to consider several design criteria in a balanced and unified way. Approximations that are typical to classical tuning rules are not needed. Evolutionary programming is often criticized for two reasons: algorithms are computationally heavy and convergence to the optimal solution can not be guaranteed. PID controller tuning is a small-scale problem, and thus computational complexity is not really an issue here. The optimization problem is defined in the paper in such way that convergence is not a problem. We are not looking for an exact solution to a complex optimization problem but instead we are looking for a group of acceptable solutions to a more loosely defined problem. Good initial values provided by some classical tuning rules also help the algorithm to find acceptable solutions rather easily.

MIMO Controller Design Using SISO Controller Design Methods

Abstract A two step procedure for the design of multivariable controllers is presented. First SISO controllers are designed for every process transfer function matrix element separately. Then the MIMO controller is computed by inverting a matrix containing the reciprocals of the SISO controllers. It is shown that complete decoupling is achieved using the controller. A modified procedure for the design of MIMO PI controllers is also given. Three examples are included.

An Expert System for the Multivariable Controller Design

Abstract In this paper a software package for the design of centralized and decentralized multivariable controllers is represented. The software package consists of numerical calculation programs and an expert system. The paper first describes the different subtasks of the multivariable controller design. Next a short descriptions of the design and analysis methods included in the software package is given. The description of the implementation of the expert system finishes the paper.

Multivariable Control Tuning with an Expert System

Abstract In this paper a software package for the design of centralized and decentralized multivariable controllers is presented. The software package consists of numerical calculation programs and an expert system. The paper first describes the general structure of the expert system. Next a design example is given. This example demonstrates how the expert system helps the user to design and tune a multivariable controller for a distillation column.

A DISTANCE LEARNING COURSE IN MODELLING AND SIMULATION

Abstract The paper describes web-based distance learning material for a course in modelling and simulation. The course covers topics like systems and models, physical modelling, system identification, and simulation. An essential part of the material are interactive demos that are implemented with Matlab, Simulink and Matlab Web Server.

Collection of Java Applets for PD Controller Design

Abstract A collection of modeling tools and PID controller tuning methods is implemented using the Java programming language and made accessible as an Internet site. The contents and structure of the site are presented. Descriptions of the modeling and tuning methods are given.

Scaled Gain Matrix -A New Method for Interaction Analysis

Abstract A new method for interaction analysis is given in this paper. The method is based on the scaling of input and output variables. Although a large gain between an input and an output indicates strong coupling, the process gain matrix can not be directly used for interaction analysis, because it depends on the scaling of input and output variables. In this new method the input and output variables are rescaled so that in the new scaled gain matrix, corresponding to the rescaled variables, the elements are directly comparable with each other.

Multivariable Pi-Controller Tuning for a Binary Distillation Column

Abstract Binary distillation columns offer a deluanding task for the design and tuning of controllers. They are highly nonlinear processes with strong interactions. In this paper four different multivariable PI controller tuning methods are compared experimentally using a pilot distillation column. The methods used are proposed by Davison (1976), Penttinen and Koivo (1980), Maciejowski (1989) and Lieslehto(1991).

An Expert System with a Hypercard Interface

Abstract In this paper a software package for the design of centralized and decentralized mult ivariable controllers is represented. The software package consists of numerical calculation programs and an pypert. system. The paper first, describes the different subtasks of the multi variable controller design. Next a short descriptions of the design and analysis methods included in the software package is given. The description of the implementation of the expert system finishes the paper