A PID autotuner utilizing GPC and constraint optimization

A solution to the PID autotuning problem is presented which involves constraining the parameters of a discrete second-order discrete-time controller. The integrator is forced into the regulator by using a CARIMA model. The discrete-time regulator parameters are calculated by optimizing a generalized predictive control criterion, and the PID structure is ensured by constraining the parameters to a feasible set defined by the discrete-time Euler approximation of the ideal continuous-time PID controller. The algorithm is extended by incorporating constraints on the amplitude and slew-rate of the control signal. Simulation studies for a system of coupled tanks have indicated that the method performs well, and that signal limitations can be included in a straightforward manner.<>