Modeling and control of hierarchical systems with fuzzy systems

Abstract

Fuzzy systems have a dual role: on the one hand, they are rule-based systems constructed from a collection of fuzzy IF-THEN rules; on the other hand, they are nonlinear mappings with nice mathematical properties like universal approximation. In this paper, we use fuzzy systems to model higher levels of hierarchical systems and design controllers for the hierarchical systems. In the modeling part, we consider three-level hierarchical systems where the lowest level comprises the plant and conventional feedback controller, the middle level performs supervisory operations to guarantee the stability of the whole system, and the top level is a planning level that provides control targets for the lower levels and communicates with the environment. The plant is modeled by differential equations, and the supervision and planning levels are modeled by fuzzy systems. Two case studies are presented: integrated planning and control of mobile robots, and intelligent vehicle/highway systems. In the control part, two types of control systems are designed for the hierarchical model, depending upon whether or not the plant model is available. In both cases, the whole hierarchical control systems are shown to be stable, with the tracking errors converging to zero.