Fuzzy-based interaction prediction approach for hierarchical control of large-scale systems

This paper presents a fuzzy-based interaction prediction approach (F-IPA) for two-level optimal control of large-scale systems. The design procedure uses a decomposition/coordination framework of hierarchical structures. At the first level, the system is decomposed into subsystems for which subproblems are formed. At the second level, a fuzzy coordinator is used to predict the coordination parameters needed to coordinate the solutions of the first level subproblems. The fuzzy coordinator uses a critic vector to evaluate its performance and learn its parameters by minimizing an energy function. The proposed control scheme is implemented on a two-degrees-of-freedom (2DOF) model of robot manipulator with parallel links and two inverted pendulums on cars. Simulation results are obtained and compared with the gradient-based interaction prediction approach (G-IPA) and a centralized optimization.