Auto-tuning method for PI controllers under Symmetric-Send-on-Delta sampling strategy

Abstract

In this work, an auto-tuning method for PI controllers applied in a loop under a Symmetric-Send-On-Delta sampling strategy is proposed. Auto-tuning algorithms usually implement an identification phase to obtain some information about the system dynamic that is used for tuning the controller. In our proposal, two frequency response points are estimated through an iterative procedure. With this information, a simple tuning rule is applied which guarantees robustness against limit cycle oscillations that can be induced by the SSOD sampler. Both the robustness measure to limit cycles and the identification procedure are based on the describing function technique, whose validity in the networked control systems under study requires a low incidence of losing packets as well as the transmission delays to be known or negligible compared to the system’s dynamic. Comparatives with other classical tuning rules support the suitability of the proposed method. The auto-tuning procedure has been validated through a simulation study reflecting its applicability for the most common dynamics found in industrial processes.