Predictor-Based Active Disturbance Rejection Control of Wet Flue Gas Desulfurization System With Delay Robustness and Simplified Tuning

Flue gas desulfurization in thermal power plants is critical to the environmental and human health. Aiming for field application, this article presents a new predictor-and-observer-based active disturbance rejection control (ADRC) scheme for Wet Flue Gas Desulfurization (WFGD) process which is subject to long time delay and unknown disturbances. First, the working principles and control problems of WFGD system are illustrated, and the WFGD process model is identified based on the field experimental data. Then, a predictor-based ADRC approach is carefully devised to address its control issues. Specifically, a filtered smith predictor (FSP) is designed to obtain stable delay-free prediction output, and a reduced-order model-assisted extended state observer (MESO) is devised to estimate the “total disturbance” with less phase lag. Based on the equivalent control structure, its stability, including the perfect delay matching and delay mismatch cases, is theoretically investigated using the zero exclusion criterion. A simple “separation tuning procedure” is also provided, where the ADRC parameters are tuned directly based on the delay-free process, while the FSP filter is specified to achieve desired robustness. Finally, the validity of the proposed method is demonstrated by comparison simulations and field tests in a 1000MW thermal power unit.
Note to Practitioners
—This article focuses on sulfur dioxide (SO
$_2$
) control and proposes a new predictor-based ADRC approach for the WFGD process subject to long time delay and unknown disturbances. Taking the WFGD system as an example, it systematically describes the whole process of modeling, control system design and simulation & experimental verification, which can provide an intuitive reference for practitioners. The zero exclusion criterion is used to analyze the closed-loop stability and determine the stable delay range for a given predictor-based ADRC control system. A detailed tuning procedure is also provided, which can simplify the time-consuming tuning process and help engineers grasp the performance and robustness of the control system. It is evident that the proposed method can be extended to a general class of industrial time-delay systems, which facilitates the practitioner’s practice in general single-input-single-output time-delay system control. The field test results demonstrate the superiority of the proposed method and show promise for practical application in other systems.