Paper FA1.7
Khandelwal, Shubham (Indian Institute of Technology Hyderabad), Detroja, Ketan (Indian Institute of Technology Hyderabad)
Detuning Iterative Continuous Cycling Based Multi-Loop PI Control for Multivariable Processes
Scheduled for presentation during the Regular Session "Linear Systems and Robust Control" (FA1), Friday, November 29, 2019,
10:15−12:45, WZ Building Room WZ416
2019 Australian & New Zealand Control Conference (ANZCC), November 27-29, 2019, Auckland, New Zealand
This information is tentative and subject to change. Compiled on May 1, 2024
|
|
Keywords Process Control & Instrumentation, Linear Systems
Abstract
Multivariable systems are universal and unavoidable in practice. Designing a PI controller for such multi-input multi-output system is challenging. In most cases, the complexity of a design method increases significantly beyond 2x2 systems. Therefore, along with effectiveness and robustness of the controller, simplicity and easy scalability of the design method are some of the key requirements from a control system design method. We propose a Detuning Iterative Continuous Cycling (DICC) method for decentralized PI control of MIMO processes. The proposed DICC design utilizes the idea of continuous cycling (CC) for obtaining the ultimate parameters (UPs) for the effective open-loop transfer functions (EOTFs). While for TITO systems the controller settings can be easily derived for the EOTFs, controller tuning for higher dimensional systems is challenging due to complicated EOTF dynamics. Therefore instead of EOTFs, it is proposed to use the effective transfer function (ETF) description of the large scale MIMO system. The ETFs are used to obtain the UPs during the closed loop CC test. Thereafter detuning based control design is proposed for obtaining multi-loop PI controller settings. The wide applicability, effectiveness, simplicity and easy scalability of the proposed DICC method has been demonstrated by considering various 2-,3- and 4- dimensional MIMO systems. Further, robustness of the proposed design has also been tested by introducing a plant-model mismatch.
|
|