Knowledge based controller selection for electrical drives with imperfect mechanic

A self-commissioning system for the high-performance speed control of electric motor drives is presented, which requires a knowledge based controller selection tool (KBCS-tool) as a basic building block. It is assumed that the mechanic can be sufficiently modeled as a two-mass system, including the nonlinearities of friction and backlash. To build up the KBCS-tool, different control concepts are investigated and the results are stored as rules in a knowledge base, which is then capable of selecting a suitable controller for a mechanical system regarding its identified structure and parameters as well as certain requirements given by the operator (e.g. dynamic, overshoot, stiffness, etc.). This paper highlights the working principle of the knowledge base, the determination of its rules and its integration into the self-commissioning system, together with the performance of different controller concepts.