【摘要】 故障预测作为一种能够实现对故障"未卜先知",进而可以提前干预的故障诊断技术,近年来受到了越来越多的关注。提出了一种基于动态内部主元分析(DiPCA)的故障预测算法,并针对田纳西-伊斯曼(TE)化工过程进行了仿真研究。与传统的主元分析(PCA)算法不同,DiPCA考虑了数据沿时间维度的动态变化,通过构建动态隐含变量及其数学模型,捕获数据中主要的动态特性,进而根据历史数据推演出系统的未来变化,更适用于系统的故障预测问题。研究表明该算法在阶跃变化、随机变化和粘滞等不同类型的故障下均能提前预报故障的发生,验证了算法的有效性。更多还原

【Abstract】 As a kind of fault diagnosis technology that can realize the unpredicted of faults and can intervene in advance,fault prediction has received more and more attention in recent years.In this paper,a fault prediction algorithm based on dynamic internal principal component analysis(DiPCA) is proposed,and the simulation of Tennessee-Eastman(TE) chemical process is carried out. Different from the traditional principal component analysis(PCA) algorithm,DiPCA considers the dynamic change of data along the time dimension.By constructing dynamic implicit variables and their mathematical models,the main dynamic characteristics of the data are captured,and then the system can be derived based on historical data.Future changes are more applicable to system failure prediction problems.The research shows that the algorithm can predict the occurrence of faults in advance under different types of faults such as step change,random variation and viscous,and verify the effectiveness of the algorithm.更多还原