Real-Time Force Control of a Grinding Process Using Unfalsification and Learning Concept

摘要

Conventional controller synthesis relies on a plant model that is not falsified by the open-loop experiments. Alternatively, without any plant model or prejudicial assumptions it is possible to find a set of controllers that are not falsified by the performance specification or the measured data (Safanov and Tsao, 1997). This approach is used to implement a real-time force controller on a reciprocating surface grinder.