Identification and Tuning Methods for PI Control Systems Based on Symmetric Send-on-delta Sampling

• Published in: International journal of control, automation, and systems Vol. 17; no. 11; pp. 2784 - 2795
• Main Authors: Moreno, José Sánchez, Losada, María Guinaldo, Visioli, Antonio, Bencomo, Sebastián Dormido
• Format: Journal Article
• Language: English
• Published: Bucheon / Seoul Institute of Control, Robotics and Systems and The Korean Institute of Electrical Engineers 01.11.2019; Springer Nature B.V; 제어·로봇·시스템학회
• Subjects: Algorithms; Closed loops; Computer simulation; Control; Control methods; Control systems design; Controllers; Engineering; Identification; Identification methods; Mechatronics; Process parameters; Proportional integral; Robotics; Transfer functions; Tuning; 제어계측공학; identification; tuning; Events; limit cycle PI controller; send-on-delta
• ISSN: 1598-6446; 2005-4092
• DOI: 10.1007/s12555-018-0911-2

Two procedures for the definition of an autotuning algorithm for event-based proportional-integral (PI) control systems are proposed in this contribution. Firstly, a method for the identification of the most common transfer functions used for tuning PI controllers is explained. The rationale of the identification method is based on the limit cycles that the event-based sampler and a convenient tuning of the PI controller can produce in the closed loop. The identification method is designed for controller tuning, that is, fitting the behaviour of the process in the range of frequencies between ω -180° and ω -135° . Secondly, a new method for tuning the event-based PI controller is explained. The main control objective of the new tuning rule is to generate a set of parameters that introduces the process in a limit cycle at a user-specified inter-event time. However, by introducing an oscillation margin in the tuning rule, the method generalizes to produce controller parameters that avoid limit cycles with a certain robustness margin. Simulations demonstrate the effectiveness of the tuning rule to force the system to oscillate at a specified frequency as a consequence of the events triggered by the event-based sampler.