Hierarchical Optimizing Control of Steady-state Large Scale Systems Under Model-reality Differences of Mixed Type — A Mutually Interacting Approach

Abstract

A new hierarchical two-level method for on-line optimizing control of stationary steady-state large scale systems is proposed. The method utilizes both the system model and the system interaction and output measurements. The available system model doesn t have to be parametric and it possesses natural parameter variables. Hence, the method is fitted into the situation usually met in the real-world when the model-reality differences are of mixed type. The coordination variables are parameters and price vector. Therefore, optimization and parameter identification mutually interact but neither of them is favoured. A notion of a point-parametric model on a set is introduced and it is shown that based on such model the method is able to cope effectively with uncertainty of mixed type.A suboptimality and an applicability of the method are investigated and a suitable coordinator strategy is proposed. The theoretical results are illustrated on an example system.