Abstract
In this study, the main contribution is a new approach to control multivariable systems by engaging an idea of hierarchical aggregation of multiple fuzzy controllers. A two-level control architecture is developed in which in addition to local fuzzy controllers focused on control of individual subsystems, a higher level controller coordinating and adjusting control actions is designed. The performance of the approach is illustrated with the use of the benchmark problem of the three-tank water control. A statistical comparison is carried where the hierarchical control strategy is compared with the one when a collection of independent individual fuzzy controllers is involved. We demonstrate that the proposed method outperforms "conventional" fuzzy control. Genetic optimization (genetic algorithm) is used in the design of the overall control architecture.