GUO Zhi.Spatiotemporal Characteristics-based Decoupling Control Method for Multimodal Operation Data of Coal Mills in Thermal Power Plants[J].POWER EQUIPMENT,2026,40(05):394-399.
GUO Zhi.Spatiotemporal Characteristics-based Decoupling Control Method for Multimodal Operation Data of Coal Mills in Thermal Power Plants[J].POWER EQUIPMENT,2026,40(05):394-399.DOI: 10.19806/j.cnki.fdsb.2026.05.010.
Spatiotemporal Characteristics-based Decoupling Control Method for Multimodal Operation Data of Coal Mills in Thermal Power Plants
To address the problem of strong spatiotemporal coupling and nonlinear time-varying characteristics in the multimodal operation data of coal mills in thermal power plants, which make it difficult for traditional single variable control strategies to meet the requirements of multivariable collaborative control and restrict the improvement of operational stability and energy efficiency, a multivariable decoupling control method that integrates spatiotemporal characteristics was proposed. By analyzing the operation principle of the coal mill and the characteristics of multi-source data, an improved adaptive wavelet threshold denoising and time-delay compensation method was adopted for preprocessing. A multi-scale temporal convolutional network was used to extract time features, and a graph attention network was used to construct a spatial topological graph to extract spatial features. Then, a dual stream cross attention mechanism was used to achieve spatiotemporal characteristics fusion. A hierarchical decoupling controller with a dynamic compensation layer and a robust control layer was designed. Multi-variable decoupling control was achieved based on the characteristic decoupling matrix. Furthermore, the verification was conducted using the coal mill of a 660 MW unit. The results show that compared with traditional proportional-integral-differential (PID) control, this method improves the overshoot of pulverized coal fineness by 72.6%, shortens the recovery time of outlet temperature by 51.1%, reduces the standard deviation of vibration intensity by 43.7%, and decreases the coupling degree of air coal ratio by 84.5%. The probability of maintaining the outlet temperature within the set value ±2 ℃ range under coal quality disturbance is 93.6%, the number of abnormal shutdowns is significantly reduced, and high control accuracy is still maintained under grinding roller wear condition. This method can effectively reduce the coupling effects of multiple variables, improve the stability and control accuracy of coal mills operation, and provide an effective technical approach for optimizing coal mills control systems in thermal power plants.
Kim D H , Lee S B , Kim Y H , et al . Pulverizer condition monitoring system using acoustic emission technique in thermal power plant [J]. Journal of the Korean Society for Nondestructive Testing , 2017 , 37 ( 6 ): 435 - 442 .
Blondeau J , Kock R , Mertens J , et al . Online monitoring of coal particle size and flow distribution in coal-fired power plants: dynamic effects of a varying mill classifier speed [J]. Applied Thermal Engineering , 2016 , 98 : 449 - 454 .
Research on Hydrodynamic Optimization Technology for Deep Peak Regulation at 30% Rated Load of 1 000 MW Ultra‑supercritical Boiler
Research on the Construction of a Digital Twin Operation and Maintenance Model for Integrated Energy Systems Based on the Source Side of Thermal Power Plants
Optimization of the Coordinated Control System in a 660 MW Ultra Supercritical Unit
Research on Architecture of Intelligent Control Systems and Auxiliary Optimization Technology for Thermal Power Plants
Research on Exhaust Gas Temperature Control and Thermal Efficiency Optimization of Waste Incinerators
Related Author
DU Chengde
GAO Peng
SHI Lin
DAI Xiaoye
XU Shuhan
YOU Xiaohui
LI Zhou
Zhang Jin
Related Institution
CHD Zouxian Power Plant
Department of Energy and Power Engineering, Tsinghua University
Beijing Guodian Zhishen Control Technology Co., Ltd.
Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education, Southeast University