基于最小窗口MPCA的间歇过程 实时监视和故障诊断软件
来源期刊:中南大学学报(自然科学版)2003年第4期
论文作者:赵立杰 柴天佑 王军延
文章页码:463 - 466
关键词:多元统计分析;多方向主元分析;过程性能;监视;故障诊断
Key words:multivariate statistical analysis; multiway principal component; process performance; monitoring; fault diagnosis
摘 要:间歇过程起停频繁,状态变化反复,其过程监视和故障诊断十分困难.实时监视和故障诊断软件在原有DCS系统软、硬件设备基础上,基于最小窗口MPCA非线性多模型建模和监视方法,采用C,VB和MATLAB语言混合编程开发,用于挖掘数据中隐含的信息,解决批次过程实时在线监视和诊断问题.将该套软件应用于某化工过程,能早期预报和诊断异常情况,为操作人员监视和评价过程性能提供了可靠的依据,提高了过程操作的安全性,同时使产品质量提高.
Abstract: Performance monitoring and fault diagnosis are difficult due to the frequently start-stop operation and the changeable state in the batch process. Based on the existent hardware and software of DCS system, online monitoring and fault diagnosis software adopts a minimum window multiway principal component method with nonlinear multi-model structure and real time monitoring mode, and captures the latent information hidden in data. The software was programmed with C, VB and Matlab languages. It was applied in a chemical process to monitor batch processes performance, and diagnosis abnormal state. The results show that the software can be used to improve operation safety and consistent production quality.