Robust Controller Design and Implementation

The students employed robust control theory to initially design a controller following the procedures detailed in [1]. Using a D-K iteration procedure, a robust 12th order controller with a structured singular value, , less than one was obtained. The controller was implemented in the real system. As one might expect with a physical system, the simulations did not precisely match reality. The nonidealities of the pumps, level sensors, and head losses in the piping all contributed to these discrepancies. Other unmodeled phenomena witnessed by the students include the formation of vortices in the upper water tanks above the drainage holes and spontaneous triggering of the level alarms due to condensation. Despite the lack of perfect agreement between theory and practice, the students were able to generate controllers with robust performance guarantees. Representative results demonstrating the disturbance rejection capability and setpoint tracking performance of one controller design is shown in Figures 7 and 8, respectively. This controller was designed for disturbance rejection, which results in excessive input moves for setpoint moves. A robustly performing setpoint tracking controller was also implemented. This design requires an additional setpoint filter in order to satisfy the constraints on the input moves. The students clearly mastered a moderately complex control problem.