Strategies for Industrial Multivariable Control: with Application to Power Plant Control

Multivariable control strategies well-suited for industrial applications are suggested. The strategies combine the practical advantages of conventional SISO control schemes and -technology with the potential of multivariable controllers. Special emphasis is put on parameter-varying systems whose dynamics and gains strongly depend upon one or more physical parameters characterizing the operating point. This class covers many industrial systems such as airplanes, ships, robots and process control systems. Power plant boilers are representatives for process control systems in general. The dynamics and gains of power plant boilers strongly depend upon the load (instantaneous power production). This non-linear behavior suggests the use of gain-scheduling. For parameter-varying systems the multivariable control strategies can hence be divided into the following subsequent steps: choice of architecture, choice of model type and parameter estimation algorithm, choice of LTI controller synthesis framework and choice of gain-scheduling approach. State-of-the-art methods are considered and new results are derived. A specific strategy is chosen for the solution of a power plant control problem.