Self-tuning state control for a poorly damped mechanical system

As compared to the tuning of a PI-controller, the commissioning of a state controller for an electric drive requires exact knowledge of the plant´s dynamics and is not easily done by hand. To facilitate tuning of a state controller, an automatic plant identification and controller optimization procedure has been developed. The identification algorithm is based on the common extended least squares algorithm. Some modifications improve the robustness of the algorithm against deterministic disturbances. The controller optimization is done by an improved vector optimization method. Fuzzy reasoning is employed to make the optimization easily configurable. Results show the great reliability of the identification and the excellent performance of the optimized controller