A two-level gradient based approach for intelligent coordination of large-scale systems. Part I

In this paper, the concept of coordination is introduced within the framework of two-level large-scale systems and a new approach based on interaction prediction principle is presented. The proposed approach is formulated in an intelligent manner in such a way that it provides a new strategy that can be used to synthesize an on-line supervisory controller for the overall two-level large-scale systems, extendable to multi-level control systems. By using the new methodology, which is based on using neural network for modeling each sub-system, typical gradient method for optimization of first-level sub-problems, and the gradient of the interaction prediction errors related to the predicted interactions, at the second level, the coordination of the overall large-scale system is done. Simulation results demonstrate the effectiveness of the proposed strategy in compare to the classical methods