超临界直流炉机组建模及模型复杂方程优化  

Modeling of supercritical once-through boiler unit with complex equations optimization

作　　者：韩程浩 杜鸣 牛玉广[1] 崔召飞 李青[1] HAN Chenghao;DU Ming;NIU Yuguang;CUI Zhaofei;LI Qing(School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China)

机构地区：[1]华北电力大学控制与计算机工程学院,北京102206

出　　处：《热力发电》2025年第1期88-98,共11页Thermal Power Generation

摘　　要：为了根据超临界直流炉机组动态特性设计直流炉机组协调控制算法,以某超临界600 MW直流炉燃煤机组为研究对象,建立该机组的机理模型。考虑到长短时记忆(long short-term memory,LSTM)神经网络算法在处理长时序列数据的优越性,对于所建机理模型中锅炉部分复杂的工质吸热方程和蒸汽压差方程以及汽轮机部分复杂的过热蒸汽流量方程和负荷系数方程,采用LSTM神经网络结合机组真实运行数据进行方程拟合,建立了能够捕获机组各状态运行特性的复杂方程优化子模型。对于所建立模型中的动态参数,结合机组的动态运行数据,采用改进的差分进化算法进行参数寻优,最终得出机组完整的状态方程模型。对于所建立的模型进行开环阶跃扰动测试和闭环运行验证,结果表明所建立的模型能够精确地反映超临界机组的动态运行特性,模型闭环运行过程中的主蒸汽压力、分离器蒸汽焓值和机组负荷的平均绝对百分比误差均小于1.76%,故所建立模型可用于超临界机组协调控制算法研究。In order to study the dynamic characteristics of supercritical once-through boiler units and design a coordinated control algorithm for it,a supercritical 600 MW coal-fired once-through boiler unit is taken as the research object and a mechanism model of the unit is established.In the model,the complex heat absorption equation and steam pressure difference equation in boiler section,as well as the superheated steam flow equation and unit load coefficient equation in turbine section are fitted by the long short-term memory(LSTM)neural network with real operational data from the unit given the algorithm's superiority in handling sequence data with long-term dependencies.This established four sub-models of the equations that can capture the operating characteristics of the unit.For the dynamic parameters in the mechanism model,an improved differential evolution algorithm is proposed to identify.After this,a complete state equation model is obtained.The established model is verified through open-loop step disturbance testing and closed-loop simulation upon historical operation data.The results show that the established model can accurately reflect the dynamic operating characteristics of the unit.The mean absolute percent error of the main steam pressure,steam enthalpy in separator and unit load are all less than 1.76%,which means the established model has high accuracy and can be used for research on coordinated control algorithms of supercritical units.

关 键 词：直流炉机组 机理建模 长短时记忆神经网络 差分进化算法 状态空间方程模型 

分 类 号：TM621.2[电气工程—电力系统及自动化]