Reconfigurable Control Using Constrained Optimization

A conceptual scheme for reconfiguring control systems in the event of major failures is advocated. This addresses a 'big problem' which is understandable by the man in the street, delivers enormous benefits if it can be made to work, and requires interesting and challenging research from control and systems theorists and practitioners. The scheme relies on the convergence of several technologies which are currently emerging: Constrained predictive control, High-fidelity modelling of complex systems , Fault detection and identification, and Model approximation and simplification. Much work is needed, both theoretical and algorithmic, to get this scheme to work, but we believe that there is enough evidence, especially from existing industrial practice, for the scheme to be considered achievable. After outlining the problem and proposed solution, the paper briefly reviews constrained predictive control, object-oriented modelling , which is an essential ingredient for practical implementation, and the prospects for automatic model simplification. The paper emphasises some emerging trends in industrial practice, as regards modelling and control of complex systems. Examples from process control and flight control are used to illustrate some of the ideas.