Explicit design of a predictive self-tuned PID controller

A novel predictive self-tuning PID (proportional-integral-derivative) controller is presented. The controller design is based on the linear prediction criteria in estimating the control variable from previous error values, providing enough time for the computations required for the self-tuning algorithm. The design procedure adopts the pole-cancellation approach and least-squares estimation in identifying the model parameters and calculating the controller coefficients using the estimated model parameters. Simulation results on different process models have shown that the controller yields improved performance and ability to cope with significant dead-time processes in comparison with conventional PID self-tuners. Specifically, the proposed scheme provides an improved performance in the sense of the classical figures of merit, such as overshoot, settling, and rise-time. The emphasis of the approach on higher-order dead-time models demonstrates its predictive capability in following the self-tuned compensation for abrupt changes and plant parameter variations