PUB - Publikationen an der Universität Bielefeld

An FPGA (Field Programmable Gate Array) implementation and suitable power electronics can lead to a fast torque response in motion drive applications. However, when the controller parameters or its structure have to be adapted to internal and external varying conditions, e.g., when a selfoptimizing control system is pursued, a static implementation might not lead to the best utilization of reconfigurable resources. This contribution outlines the implementation of a self-optimizing system composed of several possible hardware and software realizations of controllers for a permanent magnet servo motor. How well a specific controller realization is suited to the current situation is evaluated based on control quality and realization effort (i.e., CPU time, reconfigurable area). A System-on-Chip architecture is presented, which enables an on-line exchange of FPGA- and CPU-based realizations of controllers to optimize resource utilization and control quality. It is shown that by using dynamic hardware reconfiguration, such self-optimizing controller can be implemented based on FPGA technology. Furthermore, the design-flow including self-developed tools is outlined. Experimental results show that the proposed scheme works satisfactory.