Design of a fuzzy controller for a rotating oscillatory†
Torsten Bertram · International Journal of Systems Science · 1993
In the effort to automate industrial processes, we find that the controlled systems become more complex due to their size, the practical use and the desire for the greatest possible safety. We present our experience of the development of a fuzzy controller for a rotating oscillator. The physical system ‘rotating oscillator’ is an obvious choice for the derivation and examination of a fuzzy controller, because the non-linear components (friction effects in the bearings, the friction forces between the soft spring and the mechanical guidance, the non-linear characteristic of the spring and the backlash of gears) cannot be sufficiently modelled. On the basis of experimental results, we show that the fuzzy controller can be designed by making use only of heuristic knowledge although an analytical mathematical model is not available. The parameters of the controller are chosen with the help of a stochastic optimization. Additionally, we present fuzzy rules for compensation of stationary control deviations, which yield a high accuracy of the position of the rotating mass.