Robust design of smith predictor controllers under pre=prescribed contraints using genetic algorithm

摘要

In many cases, the standard process control model involving a frist order lag plus time-delay can be incorporated as the model in the Smith predictor controller (SPC). In this paper, a genetic algorithms is used to show that for this type of model, the design of optimal tuned (form minimujm ISE) proportional plus integral Smith predictor controllers when subject to a broad range of robustnes sconstraints can be simplified by choosing the integral time-constant equal to the dominant process time-constant. This result in a single tuneable parameter Smith predictor controller. (SP-SPC). Controller tuning curves are then produced which faiclitate the design of such a SPC which satisfy pre-prescribed gain margins and delay margins constraints. Finally, the results of the SP-SPC are compared to optimally tuned PI controllers when subject to the same robustness constraints. Copyright ~direct 2000 IFAC