Selftuning of classical controllers with fuzzy-logic

Selftuning of classical PI-controllers for processes without known mathematical models is achieved by applying heuristic fuzzy rules. Step changes of the reference input are used to assess control performance in terms of overshoot, rise time and setting time of the step response. Similar to the strategy of an experienced plant operator, a fuzzy logic “supervision and tuning module” can conclude from these features, which changes of the controller parameters will improve control behaviour. The concept was tested in simulation series with Matlab/Simulink. Results show the benefits of the proposed fuzzy tuning module applied to 8 different simulated test plants. Experiments with speed-regulation of a DC-servomotor indicate that the concept is suitable for processes with noisy signals as well.