Data based design of PID controllers for a magnetic levitation experiment.

摘要

Data based controller design techniques provide an excellent alternative to the traditional model-based control design methodology. They have important significance in real world control engineering, where mathematical models of the plant to be controlled are not available or accurate models are difficult to obtain. Keel and Bhattacharya demonstrated a data based approach for design and synthesis of PID controllers. The technique introduced assumes ideal conditions such as perfect linear time invariant systems and availability of frequency response at all frequencies. In this project, we test the technique under a real world environment where the assumptions no longer hold. Using the data based controller design technique, we design PID controllers for a magnetic levitation experiment and validate the design method for practical use. The frequency response experimental data was collected for a finite range of frequencies that is appropriate for the system response. Using this frequency response data, the entire set of stabilizing PID controllers is determined. From this stabilizing set, PID gains are computed to satisfy multiple performance specifications and the corresponding controller is implemented to experimentally validate and illustrate the performance of the closed-loop system.