Tuning of fuzzy controllers - an overview

摘要

There are two ways to design fuzzy controllers, by using mathematical models or linguistic models. In this second case the information about structure and parameters of the controller comes from the experience of the process engineers allowing to find the controller structure and the rule base, and from starting-up to get the adequate parameters of the controller. Tuning rules from a relation between the actual situation of the controlled process and the necessary manipulations for a better behaviour. Therefore tuning is an on-line optimisation at the real plant with limitations. The fuzzy-unit is the basic module for fuzzy controllers and more complicated fuzzy control structures with a high number of degrees of freedom. The number of parameters, which are to tune, should be as small as possible. This contribution describes some methods for a successful tuning of fuzzy controllers under different conditions. regarding the real process as well as possible. For tuning fuzzy units, autonomous systems, linear methods, large scale linearisation and automatic tuning is regarded as an overview.