From dynamic data to fuzzy state-space controllers: Methodology and applications

This contribution presents a closed methodology to derive fuzzy state space controllers from dynamic process data: Sugeno-type fuzzy models with multivariate membership functions in I/O representation are identified by means of fuzzy clustering, LS and optimization methods. An equivalent fuzzy state-space representation is derived. Employing that a fuzzy state-space controller is desgined. To compensate for steady state errors an adaptive set point filter is calculated. The concept is applied in two case studies including an industrial hydraulic linear drive.