【摘要】 主元分析(PrincipalComponentAnalysis,PCA)已广泛应用于复杂工业过程的运行状态监控。然而,传统的PCA方法仅构造了生产过程的静态线性关系,无法从根本上有效处理具有较强动态特性的实际工业生产过程。动态主元分析(DynamicPCA,DPCA)是一种将传统PCA分析推广到动态多变量过程的方法,但其较大的计算负荷阻碍了其实际应用。本文对文献中的DPCA作了算法上的简化,减少了实施中的计算量,并将其应用于重油分馏塔的动态运行故障监测与诊断。研究结果表明了方法的有效性。更多还原

【Abstract】 Principal Component Analysis(PCA) has been widely used for monitoring complex industrial process. However,the traditional PCA only constracts linear static relations among the process variables, it cann’t fundamentally and effectively deal with the real industrial process which possesses strong dynamic characteristic. Dynamic Principal Component Analysis (DPCA) is a method by extending static PCA to monitor dynamic multivariate process. However, the large computing load of DPCA limits it’s application to industrial process. The paper predigest the arithmetic of the DPCA in literatures, reducing the real computing work. This method is used in fault detection and diagnosis of the typical Heavy Oil Fractionator process, and the results verify the effectiveness of the improved DPCA.更多还原