摘要
锅炉过热汽温是燃煤机组运行中的重要参数,过高、过低都会对机组的安全经济运行构成威胁。由于锅炉结构复杂,系统庞大,汽温对象具有变时滞、变参数等特性,喷水减温系统采用的串级PID控制,在大范围变工况下效果往往很不理想,且PID参数整定耗时耗力。为此,该文针对600 MW超临界锅炉过热器的喷水减温系统,研究了过热汽温神经网络(ANN)内模控制方案。基于Matlab建立了汽温系统的ANN正模型和逆模型,并设计出ANN内模实时控制器。仿真表明,与原串级PID控制相比,该方案显著改善了过热汽温的控制品质。
Boiler SST was an important parameter closely related to the safe and economic operation of a coal-fired power plant. Superheated steam temperature either too high or too low will pose a threat on the operation safety and economy. Because the boiler superheater system was relatively complex with large delay,large inertia and higher non- linearity,satisfaetory control effects can often not be obtained with conventional cascade PID control scheme at wide operating range. Retuning of the PID parameters was often a time-consuming and labor-intensive work. For this rea- son,a neural network IMC scheme was studied for the two-stage desuperheating system of a 600 MW supercritieal boiler unit. studying the system with the method based on superheated steam temperature control scheme. The neural network direct models and inverse models of the superheater system are built and trained with MATLAB software. Then neural network model based IMC controllers for the superheated steam temperature were designed,programmed and tested with a full-scope power plant simulator. It was shown that the neural network IMC scheme can significantly improve the superheated steam temperature control quality compared to the original cascade PID control.
出处
《自动化与仪表》
北大核心
2013年第4期10-13,共4页
Automation & Instrumentation
基金
国家自然科学基金项目(61174111)
中央高校基本科研业务费专项资金资助
关键词
超临界锅炉
过热汽温
内模控制
人工神经网络
supercritical boiler
superheated steam temperature (SST)
internal model control(IMC)
artificial neural network ( ANN )