Operator Interaction and Optimization in Control Systems

Författare

Johan Åkesson

Summary, in English

The two themes of this thesis are operator interaction and optimization, primarily with application to complex plants in the process industry. Four important aspects that should be considered in the control system design for such plants are identified; safety, performance, operator interaction and operator acceptance. In particular, operator acceptance is considered as an aspect of major importance. Since the operators are usually responsible for the operation of the plant, it is essential that the control system is well understood and accepted by the operators. The inverted pendulum serves as introduction and motivating example of this thesis. In this application, the trade-off between safety and performance is important, in combination with close operator interaction. A frame work for grade changes is presented. The proposed method integrates the use of a tool for sequential control, JGrafchart, and a method for dynamic optimization. It is shown how a simple parameterization of the control variable trajectories may be implemented and executed as a Grafcet sequence. Model Predictive Control (MPC) is an optimization-based control strategy which has won widespread use in the process industry. In this thesis, tools for simulation and analysis of an MPC controller for linear systems are presented. The tools are used to investigate the effects of computational delay, and a method for delay reduction is proposed. A method for ensuring offset-free tracking in centralized control of MIMO-systems i presented. It is shown how a particular choice of disturbance model can be used to prove that an observer-based controller will contain integral action. Specifically, the case where the number of measured variables exceeds the number of inputs is treated.

Avdelning/ar