Abstract:

In complicated industrial systems, without the first component model and just by use of routine data, factor analysis can model the process operation and take full consideration of general sense of model error, therefore it has a good application prospect.Aimed at the dynamic characteristics in the actual process, the paper proposes a novel method based on dynamic factor analysis (DFA) to model the process data.In DFA, the sample matrix is extended based on the auto-regressive (AR) model, so it can well extract the information of auto-correlation and cross-correlation among process variables.In order to evaluate the operational condition of the process, the paper constructs several statistics as the monitoring indices to measure the features and errors, respectively.The application in the Tennessee-Eastman (TE) process shows the superiority of the DFA-based method.

Key words:

动态因子分析;过程监控;建模, 监控指标

摘要：

在复杂工业系统的监控中，因子分析（FA）方法不需要专业的机理知识，应用系统日常运行数据建立模型，充分考虑了模型误差的普遍意义，具有较大的推广价值。针对实际过程的动态特性，基于自回归（AR）方式扩展过程变量数据矩阵，本文提出一种动态因子分析（DFA）的数据建模方法，充分提取了变量的自相关信息和互相关信息。另一方面，将DFA引入过程监控中，构建统计量作为监控指标，分别衡量变量的特征信息和误差信息，从而实现对动态过程运行状况的监控与评估。在Tennessee-Eastman（TE）过程中的应用研究，反映了这种方法的优越性。

关键词:

动态因子分析;过程监控;建模, 监控指标