双线性数据校正的多组分过程容错模型研究  

Fault-tolerant modeling approach based on bilinear data reconciliation

作　　者：陈嘹[1] 冯毅萍[1] 荣冈[1] 

机构地区：[1]浙江大学智能系统与控制研究所,浙江杭州310027

出　　处：《计算机与应用化学》2011年第7期807-810,820,共5页Computers and Applied Chemistry

基　　金：国家高技术研究发展计划项目(2009AA04ZX1485932)

摘　　要：可靠的测量数据是化工过程建模的关键。流程工业中的变量测量值不可避免带有误差。如果这些带有噪声误差的测量数据直接作为模型辨识的样本数据,由此得到的模型势必和真实流程模型大相径庭。本文提出了一种基于双线性数据校正的多组分过程容错建模方法,利用双线性数据校正方法来对测量数据进行预处理,然后再将协调结果和其他的测量数据一起作为模型辨识的输入,从而减少测量数据误差对于过程建模的影响。等离子裂解煤制乙炔气体分离工段的仿真结果证明了该方法的优越性。Reliable process data are the key to the modeling of chemical processes. Measurements of process variables are subjected to errors. If these data with noise are used directly as the identifying data set, the empirical model constructed will be quite different from the dynamics of real process. In this paper, a fault-tolerant modeling approach based on bilinear data reconciliation is proposed. Bilinear data reconciliation algorithm is used as a preprocessing module to reduce the measurement error. The reconciliation results together with other measured data are then fed in as model inputs. Thus, the constructed model is fault-tolerant to measurement errors. Such ability is demonstrated by simulation result of pyrolysis gas separation section of acetylene from coal by plasma pyrolysis process..

关 键 词：过程模型 容错模型 双线性数据校正 等离子裂解煤制乙炔 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TP319[自动化与计算机技术—控制科学与工程]