On Model-reality Differences in Integrated System Optimisation and Parameter Estimation

Abstract

One of the difficulties in steady-state adaptive model-based optimising control techniques is deciding whether the model to be employed will, in some sense, be faithful in its representation of the system Often the model and the system outputs can be found to be in close agreement throughout an optimising procedure and yet the efficiency of achieving the optimum operating conditions is poor This then must mean that, in optimising control, there are some fundamental differences in the model and system structures, and that the model-reality differences are defined by factors other than simply the outputs. This paper looks into these differences, and looks at how these differences can be compensated for in solving the optimising control problem.