A dynamic hybrid model of supercritical once-through boiler-turbine unit including recirculation mode and once-through mode

• Published in: Energy (Oxford) Vol. 309; p. 132941
• Main Authors: Xie, Yan, Liu, Ji-zhen, Zeng, De-liang, Hu, Yong, Li, Rui-lian, Zhu, Yan-song
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.11.2024
• Subjects: Dynamic hybrid model; INFO; INFO-XGBoost; Once-through mode; Recirculation mode; Supercritical once-through boiler-turbine unit; Dynamic hybrid model; INFO-XGBoost; INFO; Recirculation mode; Supercritical once-through boiler-turbine unit; Once-through mode
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2024.132941

Supercritical once-through boiler-turbine units play a paramount role in maintaining grid stability owing to their exceptional efficiency and flexibility. Nevertheless, achieving precise modeling of this device under various operational conditions remains a significant challenge. This study conducts modeling and simulation of an actual supercritical once-through coal-fired power plant. The findings signify that the hybrid model proposed herein adeptly encompasses all operational modes of supercritical once-through units, including start-up recirculation mode and once-through mode. To bolster model precision, an optimization algorithm grounded on weighted mean of vectors (INFO) has been employed for parameter identification, while the INFO-XGBoost was utilized to fashion a machine learning model for static parameters. It has been validated that the developed model accurately captures the dynamic characteristics of the unit. Furthermore, the utilization of the INFO-XGboost to model static parameters has been proved to enable the errors in principal outputs of coordinated control system to be within 0.35 % of MAPE across the entire operational process, thus facilitating the design of more refined and intelligent control systems. •Recirculation mode and once-through mode are covered in established hybrid model.•The main model outputs are within 0.35 % of MAPE and validated in an 1000 MW unit.•Algorithm based on weighted mean of vectors is used for parameter identification.•INFO-XGboost is considered to establish the models of static parameters.