摘要
针对一类可以采用一阶惯性加纯滞后模型描述不同工况下动态特性且随工况变化的工业过程,提出一种基于对象特征参数极大极小值的多模型集建立方法,采用递推贝叶斯概率加权方法获得全局预测模型,并以此设计多模型预测控制器以满足工况大范围变化的控制要求,同时在进行误差校正时,预先补偿由于工况动态变化所带来的模型预测误差,以提高预测精度。对电站锅炉主汽温系统的仿真结果表明在各工况下均有很好的定值跟踪能力,在大范围工况变化时,能够将主汽温度稳定在设定值附近。
Concerning a kind of industrial processes for which first-order inertia plus a pure lagging model can be used to describe their dynamic characteristics under different operating conditions and which change with operating conditions,a method was presented for setting up a multi-model set based on the maximum and minimum values of the characteristic parameters of an object. A recursive Bayesian probability weighting method was used to obtain an overall predictive model. On this basis,a multi-model predictive controller was designed to meet the control requirement for the operating conditions varying in a wide range. In the meanwhile,when a rectification of errors is being performed,the prediction error of the model resulting from any dynamic change of the operating condition can be compensated in advance to enhance prediction accuracy. The simulation calculation results of a utility boiler main steam temperature system show that the method under discussion enjoys a superior ability to track a set value under various operating conditions. When the operating conditions change in a wide range,it is possible to stabilize the main steam temperature near a set value.
出处
《热能动力工程》
EI
CAS
CSCD
北大核心
2008年第4期395-398,共4页
Journal of Engineering for Thermal Energy and Power
基金
国家自然科学基金资助项目(50576022)
关键词
主汽温系统
多模型集
多模型预测控制
贝叶斯概率加权
动态前馈
main steam temperature system,multi-model set,multi-model prediction control,Bayesian probability weighting,dynamic feedforward
