论文部分内容阅读
随着现代数据采集系统在生产中的应用,通常需要同时监控多个相关的过程变量,并且在化工等生产过程中,多元过程数据还会呈现自相关。已有的多元自相关过程控制方法大多仅能检出过程偏移却无法诊断出哪个变量或哪些变量组合会导致过程失控。本文针对多元自相关过程,提出了基于支持向量机(SVM)的过程在线监控和诊断方法。通过构建过程监控和偏移诊断两个分类器,可以对生产中的数据进行在线监控和诊断,基于matlab的仿真结果表明,提出的方法具有更好的监控性能和更高的诊断正确率。
With the application of modern data acquisition system in production, it is usually necessary to monitor a plurality of related process variables at the same time, and in the production of chemical industry, the multi-process data also exhibit auto-correlation. The existing methods of multivariate autocorrelation process control can only detect the process offset but can not diagnose which variable or which combination of variables leads to the process being out of control. In this paper, a multi-autocorrelation process is proposed based on support vector machine (SVM) process online monitoring and diagnosis methods. Through the construction of process monitoring and offset diagnosis of two classifiers, the data in production can be monitored online and diagnosed. The simulation results based on matlab show that the proposed method has better monitoring performance and higher diagnostic accuracy.