基于SDAE的超临界机组协调系统特性建模
Modeling of Characteristics of Supercritical Unit Coordination System Based on SDAE
马良玉 1宋子滕 1王佳聪 1孙佳明1
作者信息
- 1. 华北电力大学控制与计算机工程学院,河北 保定 071003
- 折叠
摘要
为采用预测控制方法改善超临界机组的协调控制品质,采用堆叠降噪自编码器(SDAE)建立了单元机组协调系统特性的数学模型.首先,对600MW火电机组的变工况运行数据进行归一化处理,并运用预处理后的数据完成模型的离线训练.将模型并入仿真机进行单一输入变量阶跃扰动和大幅变负荷实验来验证其动态性能.实验结果表明,上述模型能够很好地描述燃料量、汽轮机调门开度、给水量与主蒸汽压力、负荷间的复杂非线性关系,泛化能力强、精度高,可以满足超临界机组智能控制器的设计要求.
Abstract
In order to improve the coordinated control quality with predictive optimal control method,the stack de-noising autoencoder(SDAE)is employed to build the mathematical model of the coordination system characteristics for a supercritical power unit.Firstly,the simulation data obtained by the full-scope simulator of a 600 MW supercrit-ical power unit are normalized,and the pretreated data are used to complete the training of the model.Finally,the model was incorporated into the simulator to verify its dynamic performance by single input-variable step disturbance tests and large-scope variable load experiments.The simulation results show that the model is with strong generaliza-tion ability and high precision,which can well describe the complex nonlinear dynamic characteristics between fuel flowrate,steam turbine regulating valve opening and feedwater flow and main steam pressure,unit load.It can meet the design requirements of an intelligent controller of the supercritical power unit.
关键词
超临界机组/协调系统/神经网络建模/在线验证Key words
Supercritical unit/Coordination system/Neural network modeling/Online verification引用本文复制引用
出版年
2024
NETL
NSTL
万方数据