摘要
常规的锅炉负荷控制方案无法解决锅炉燃烧系统的大滞后、大延时、非线性、多扰动等特性带来的问题,基于传统的机理建模和小数据校核等方式,难以建立精确的锅炉燃烧系统模型。通过深度学习+机理建模相融合的方式,综合利用深度学习的大数据能力及机理模型的泛化能力,通过构建高精度可在线更新的锅炉性能指标预测模型,模拟锅炉燃烧过程,实现对锅炉关键工艺参数进行超前预测。基于预测模型,以锅炉热效率和NOx排放质量浓度(SCR入口)为优化目标,结合先进的燃烧优化算法,对燃烧关键控制参数进行在线优化,通过调优一次风、二次风、燃尽风等配风方式,辅助锅炉燃烧调整试验、温度场可视化建模监测,为运行人员提供不同工况下锅炉燃烧调整优化指导建议,在排放达标前提下最大程度提升锅炉燃烧效率,为机组经济性运行提供优化指导。
Conventional boiler load control schemes are unable to solve the problems caused by the characteristics of boiler combustion system,such as large hysteresis,large delay,nonlinearity,and multiple disturbances,and it is difficult to establish an accurate model of boiler combustion system based on traditional mechanism modelling and small data calibration.By adopting the fusion of deep learning and mechanistic modelling,and comprehensively utilizing the big data capability of deep learning and the generalization capability of mechanistic modelling,we simulate the boiler combustion process by constructing a high-precision boiler performance index prediction model that can be updated online,so as to realize the over-advance prediction of the key boiler process parameters.Based on the prediction model,with the boiler thermal efficiency and NOx emission quality concentration(SCR inlet)as the optimization targets,combined with advanced combustion optimization algorithms,the key control parameters of combustion are optimized online,and through the optimization of air distribution methods such as primary,secondary,and exhaust air,the boiler combustion adjustment test and the visual modelling and monitoring of the temperature field are assisted,so as to provide the operators with guidance on the optimization of the combustion adjustment of the boiler in different working conditions.
作者
冯锋
Feng Feng(Shanxi Huarentong Power Technology Co.,Ltd.,Taiyuan Shanxi 030000,China)
出处
《现代工业经济和信息化》
2025年第3期84-87,共4页
Modern Industrial Economy and Informationization
关键词
锅炉燃烧
深度学习
机理模型
融合建模
优化决策
boiler combustion
deep learning
mechanism model
fusion modelling
optimization decision-making
