Author(s)

Honghui Liao, Hong Xiao, Guangsi Xiong

Corresponding Author

Honghui Liao

Abstract

The boiler soot blowing practices is the foundation for ensuring normal boiler operation and saving coal consumption. However, improper soot blowing strategies adopted by many power plants lead to decreased boiler efficiency and steam wastage. An effective ways to address this issue is to accurately monitor ash fouling and soot blowing in real-time. This paper investigates a comprehensive approach for monitoring ash fouling on the heating surface of boiler power plants and proposes an optimized soot blowing mechanism. Considering the time-series, variability, and complexity of the boiler heating surface data, an ensemble learning model is designed to monitor the cleanliness coefficient of the heating surface and thereby a boiler soot blowing optimization control strategy is proposed. This strategy combines the characteristics of steam sootblower in reality, using main steam flow rate as an environmental feedback variable, and designs a "virtual-real" dynamic reward mechanism. Experimental results demonstrate that the model can dynamically adjust control strategies based on real-time environmental feedback, ensuring higher net heat benefits and lower soot blowing costs.

Keywords

Ash fouling; Cleanliness coefficient; Soot-blow optimization; TD3 algorithm