An investigation on data mining and operating optimization for wet flue gas desulfurization systems

Abstract

Finding the best conditions for current operation from historical data is of great meaning for power plants. In order to obtain the optimal conditions, a comprehensive evaluation criterion, using the minimum cost as an objective function, was established for a wet flue gas desulfurization (WFGD) system in this paper. A basic procedure was presented to set up the database of the system operating target conditions. To improve the accuracy of mining target data, an improved fuzzy clustering (IFC) algorithm was proposed. This algorithm used K-means results as initial conditions and fuzzy C-means algorithm as analytical method. Besides, the information entropy was also utilized in the IFC algorithm as an evaluation index. Results indicated that the proposed algorithm was more accurate than typical K-means and fuzzy C-means in data clustering. Additionally, the operating data for the WFGD system of a 600 MW unit were selected as the modeling samples. In this model, the unit SO2 removal cost was selected as the criterion for the assessment of operating conditions. The operating condition data were divided into different operating clusters based on the unit load and inlet SO2 concentration of the WFGD system. Using pH, liquid-gas ratio, and slurry density as an initial condition, optimal steady-state operating data were obtained. Finally, an overall operation database of this system was established, which could successfully obtain the continuous optimal operating conditions and provide operating guidance.