直流锅炉超超临界机组协调系统多变量预测控制方法  

Multivariable predictive control method for coordinated control system in ultra supercritical units equipped with once-through boiler

作　　者：王富强[1,2] 李晓理[2] 张秋生 张金营 

机构地区：[1]神华国华(北京)电力研究院技术研究中心,北京100025 [2]北京科技大学自动化学院,北京100083

出　　处：《热力发电》2016年第4期48-53,共6页Thermal Power Generation

基　　金：国家自然科学基金项目(61473034);高等学校博士学科点专项科研基金(博导类)(20130006110008)

摘　　要：在分析直流锅炉超超临界机组协调控制系统被控对象特点的基础上,得到被控对象传递函数数学结构形式,建立3输入-3输出模型,并将其简化为2输入-2输出模型。采用约束条件下的多变量预测控制算法,提高协调控制系统的控制品质;采用Block模块对约束条件进行简化,提高预测控制算法的计算速度;采用基于滚动窗口的最小二乘法对协调控制系统被控对象实现在线辨识,提高预测控制模型的精度。依据模糊规则建立机组协调控制系统,并以其为非线性研究对象进行了滑压模式下定负荷和变负荷仿真试验。结果表明：采用多变量预测控制算法的控制品质优于传统PID控制,且计算速度是传统预测控制算法的5~6倍。The coordinated control system in once-through boilers has such characteristics as strong coupling,time-varying and difficult control.Against these problems,the multivariable predictive control with constraint conditions was applied to improve the control quality.In this system,the block algorithm was used to simplify the constraint conditions and reduce the computing amount,thus to improve the computing speed of predictive control system.The least square method based on the rolling window was employed to realize online identification of the coordinated system＇s object,to improve the accuracy of the predictive control.The unit＇s coordinated control system was established according to the fuzzy rules.Moreover,this coordinated control system was taken as the nonlinear object to carry out simulation experiments with constant and variable load,in sliding pressure mode.The results show that,this control method obviously improves the control quality and the solution speed of predictive control algorithm.The load control quality meets the inspection requirements,which is superior to the conventional PID control,and its computing speed is obviously faster than that of the conventional algorithm.

关 键 词：直流锅炉 超超临界机组 协调控制 多变量预测控制 在线辨识 最小二乘法 非线性系统 

分 类 号：TP273.1[自动化与计算机技术—检测技术与自动化装置] TK323[自动化与计算机技术—控制科学与工程]