FA监控方法改进及其应用.pdfVIP

FA 监控方法的改进及其应用 尹雪岩，刘飞 （江南大学自动化研究所,江苏 无锡，214122） 摘要：因子分析(FA )在过程监控领域的应用近年来受到了广泛关注。其可以在确定因子以及噪声的概率密度函数后，建 立过程的生成模型。本文对FA 在过程监控中的步骤作了改进。首先对过程数据进行KMO 检验,判断检测过程是否适宜应 用FA 进行统计监控。进而，采用 Kaiser 准则确定因子个数，避免对因子解释率的计算，减少了工作量。将改进的FA 监 控方法应用于化工吸附分离过程，并于PPCA 的监控效果作对比，验证其有效性及优越性。 关键词：因子分析；KMO 检验； Kaiser 准则；过程监控 中图分类号：TP206.3 文献标识码：A A Improved Monitoring Method Based On Factor Analysis And Its Application In Chemical Separation Process YIN Xue-Yan LIU Fei （Institute of Automation, Jiangnan University, Wuxi, 214122, Jiangsu, China ） Abstract: In the past few years, the application based on factor analysis in processing monitoring is widely concerned. The generated model can be constructed after the probability density functions of factor and error were determined. The approach in process monitoring based on Factor Analysis is improved in this paper. Firstly, the process is tested by Kaiser-Meyer-Olkin approach to estimate whether the process is suitable for factor analysis. Second, The number of factors are determined by Kaiser rule, this approach reduces the workload because it doesn’t computing the explanation ratio of the factors. At last, the improved monitoring method based on Factor Analysis is applied in the chemical separation process and compared with the method based on probabilistic principal component analysis ,the validity and superiority are proved. Key words: factor analysis；Kaiser-Meyer-Olkin test ；Kaiser rule ；process monitoring 引 言 反映数据本身特征，最近被引入过程监控领域[3] [4] 过程监控是保证工业系统运行过程安全可靠 。 的有效手段，而现代工业过程具有动态复杂、变量