摘要
火力发电厂厂级监控信息系统(SIS)作为一种以优化机组运行、提高运行经济性为主要目标的信息系统,已在火电厂厂级管理方面发挥了重要作用。为了更加有效地利用SIS平台上的大量数据,提出了基于SIS平台的动态矩阵预测控制。以湖北某电厂SIS为例,将锅炉过热蒸汽温度作为控制对象,蒸汽流量作为扰动量来反映锅炉运行工况的变化。采用神经网络辨识方法建立被控对象的动态模型并将动态矩阵控制算法应用到蒸汽温度控制系统。仿真结果表明,基于仿真数据建立的过热蒸汽温度神经网络动态模型,能够很好地预测被控对象在不同运行工况下的动态特性,基于神经网络模型的动态矩阵控制策略优于传统PID的控制效果。
The plant - level supervisory information system（SIS） in thermal power plant, as an information system taking optimization of unit operation and enhancement of economic efficiency in operation as the main targets, has played an important role in plant - level management in power plants. In order to more effectively utilize the vast amount of data on the SIS platform, an advanced control strategy, i.e. dynamic matrix predictive control, has been put forward. Taking SIS of one power plant in Hubei Province as exam- ple, and making temperature of superheated steam as the controlled object, and the flow rate of steam as the disturbance variable the variation of boiler＇s operating condition has been manifested. A dynamic matrix model has been established by using the neural network recognition method, and the dynamic matrix control algorithm being applied to the steam temperature control system. Result of emulation shows that the neural network dynamic model of superheated steam established on the basis of emulated data can perfectly predict dynamic property of the controlled object under different operating conditions. The dynamic matrix control strategy based on neural network model can provide better control effectiveness than that of traditional PID control.
出处
《热力发电》
CAS
北大核心
2008年第1期58-64,共7页
Thermal Power Generation
关键词
火电厂
SIS
过热蒸汽温度
动态矩阵控制算法
神经网络
预测控制
thermal power plant
SIS
superheated steam temperature
dynamic matrix control algorithm
neural network
predictive control
