基于IMPC的燃煤发电机组多变量解耦控制策略  

IMPC-based Multivariable Decoupling Control Strategy for Coal-fired Power Generation Units

作　　者：周心睿 安硕 王兵 谢宏星 朱玉林 赵凯 王正兵 ZHOU Xinrui;AN Shuo;WANG Bing;XIE Hongxing;ZHU Yulin;ZHAO Kai;WANG Zhengbing(School of Electrical and Information Engineering,Anhui University of Technology,Ma'anshan 243032,Anhui Province,China;Xiangtan Iron and Steel Group Co.,Ltd.,Xiangtan 411100,Hunan Province,China)

机构地区：[1]安徽工业大学电气与信息工程学院,安徽马鞍山243032 [2]湘潭钢铁集团有限公司,湖南湘潭411100

出　　处：《动力工程学报》2025年第2期263-274,共12页Journal of Chinese Society of Power Engineering

基　　金：安徽高校自然科学研究重大资助项目(KJ2021ZD0042);芜湖市重点研发与成果转化项目(2023yf017)。

摘　　要：针对燃煤发电机组燃烧系统的大时滞和参数时变等复杂动态特性,以及主蒸汽压力和炉膛负压间的强耦合关系,提出一种改进多变量解耦模型预测控制(Improved multivariable decoupling model predictive control,IMDMPC)。根据机组历史运行数据,利用蜻蜓算法优化偏最小二乘法,构建以主蒸汽压力和炉膛负压为被控量、燃煤量和引风量为控制量的多变量耦合数学模型,运用多变量动态解耦算法对其进行解耦,设计多变量模型预测控制器,按照设定目标函数得到当前时刻控制量,同时实时采集系统输出值对控制器进行反馈校正。仿真实验结果表明:改进多变量解耦模型预测控制与模糊自适应PID控制和未解耦模型预测控制相比,主蒸汽压力和炉膛负压的超调量分别减少20.3%和8.6%,调节时间分别缩短2.6和29.55 s,具有更好的控制精度和鲁棒性。现场应用结果表明:改进多变量解耦模型预测控制策略能够使主蒸汽压力和炉膛负压分别稳定控制在±0.24 MPa和±0.27 Pa,提高了系统的稳定性且波动幅度明显减小,符合工业现场生产要求。An improved multivariable decoupling model predictive control(IMDMPC)was proposed for coal-fired power generation units with complex dynamic characteristics such as large time delay and time-varying parameters,and strong coupling between main steam pressure and furnace negative pressure.Based on the historical operating data of the unit,dragonfly algorithm was used to optimize the partial least squares method to construct a multivariable coupled mathematical model with main steam pressure and furnace negative pressure as controlled variables,coal consumption and induced air volume as controlling variables.The multivariable dynamic decoupling algorithm was used to decouple it,and a multivariable model predictive controller was designed to obtain the current control amount according to the set objective function,simultaneously collect real-time system output values for feedback correction of the controller.Simulation experimental results show that compared with fuzzy adaptive PID control and uncoupled model predictive control,the improved multivariable decoupling model predictive control reduces the overshoot of main steam pressure and furnace negative pressure by 20.3%and 8.6%,respectively,and the adjustment time is shortened by 2.6 and 29.55 s,respectively,with better control accuracy and robustness.On-site application results show that the improved multivariable decoupling model predictive control strategy can stably control the main steam pressure and furnace negative pressure at±0.24 MPa and±0.27 Pa,respectively,improving the stability of system and significantly reducing the fluctuation amplitude,meeting the requirements of industrial on-site production.

关 键 词：改进多变量解耦模型 预测控制 火电机组 偏最小二乘法 蜻蜓算法 

分 类 号：TM621[电气工程—电力系统及自动化]