Computational Intelligence and Automated Methods for Control Fuzzy System Design

摘要

This paper aims to present a complex method of computational intelligence for a control fuzzy system design in a situation when there is no much of prior knowledge. Initial values are obtained by rule-based computations using available body of knowledge. Two fuzzy logic automated methods, batch and recursive least squares, are used to continue the computation and modelling of the system, to build and enhance the knowledge base. The extension principle methods are used in complex environments with both, discretized and continuous functions, with crisp and fuzzy data and transformations. The application of this model is illustrated by solving the autonomous cruise control problem, specifically, the throttle system. This computational method has an advantage comparing to artificial neural networks because the later do modeling of the system based only on learning data without knowing the nature of modelling applications.