Dynamic Optimization for Controller Tuning with Embedded Safety and Response Quality Measures

Controller tuning is needed to select the optimum response for the controlled process. This work presents a new tuning procedure of PID controllers with safety and response quality measures on a non-linear process model by optimization procedure, with a demonstration of two tanks in series. The model was developed to include safety constraints in the form of path constraints. The model was then solved with a new optimization solver, NLPOPT1, which uses a primal-dual interior point method with a novel non-monotone line search procedure with discretized penalty parameters. This procedure generated a grid of optimal PID tuning parameters for various switching of teadystates to be used as a predictor of PID tunings for arbitrary transitions. The interpolation of tuning parameters between the available parameters was found to be capable to produce state profiles with no violation on the safety measures, while maintaining the quality of the solution with the final set points targeted achievable