中文摘要：

空分过程异常工况的在线诊断对于保证空分产品质量、降低＂氮塞＂等故障的发生率和减小故障损失有着重要的实用价值。主元分析（PCA）是1种数据驱动的统计建模方法,已广泛应用于复杂工业过程的运行状态监控和故障诊断,然而,传统的PCA方法不能够反应数据的动态特性。动态主元分析（DPCA）作为1种将传统PCA推广到多变量动态过程的方法,其时滞长度的确定是DPCA的关键点。本文应用动态主元分析（DPCA）方法建立了空分过程异常工况的在线诊断系统,并结合空分过程的故障诊断特性,对DPCA中时滞长度提出了可行的确定方法。实际运行效果表明该系统对故障的报警率为100%,误报率约4%,证明了文中所建立的诊断系统的有效性及文中所提出的对于复杂连续生产系统确定DPCA时滞长度的方法的有效性及可行性。

英文摘要：

On-line diagnosis of abnormal conditions of the air separation process is of great practical value to ensure product qualities,reduce the incidence and the loss of failures,such as ＇Nitrogen blockage＇.As a data-driven modeling method,Principle Component Analysis（PCA） has been widely used for complex industrial process monitoring and fault diagnosing.Since the static PCA can not catch the dynamic information of the real industrial data,Dynamic Principal Component Analysis（DPCA） is introduced as an extension of PCA for dealing with multivariate dynamic data.Determining the necessary number of time lags is an important step in DPCA.This paper applies the DPCA on the air separation process,building an abnormal condition on-line diagnosing system. Dealing with the characteristics of fault diagnosis of air separation process,a method for determining the time lags is introduced.The actual operating results,in which the warn-rate of the diagnosing system to the faults of the process is 100%and the false-warn rate is about 4%,proved the validity of the proposed diagnosing system and the effectiveness and feasibility of the proposed method of determining the time lags of DPCA.