Multi-objective design of controllers with fuzzy logic using the control engineering computation environment ANDECS

Abstract

The fuzzy-control methodology can be seen as a systematic way to specify controllers with nonlinear gain surfaces. To properly tune such controllers, computer-aided techniques can be applied provided appropriate plant models are available such as in more classical design approaches. In particular, it is shown how a multi-models/multi-criteria design for controllers with fuzzy logic can be carried out by using the ANDECS-MOPS Multi-Objective Programming System. For that a Control Data Object 'Fuzzy-Controller' is developed together with a graphical editor for defining the linguistic variables. This represents the dynamic synthesis part of the design framework. Using a model building environment like Dymola or ACSL which is linked to ANDECS via DSblock, the fuzzy controller connected to dynamic plant models can be optimized by MOPS. This is demonstrated for a multi-criteria design problem for an aerodynamically unstable aircraft.<>