基于KPCA-FCM的工业过程故障检测与诊断

发布时间：2018-07-28 12:11

【摘要】：有效的工业过程故障诊断方法,不仅能够在第一时间对出现的故障进行识别和处理,避免可能出现的危险,同时也要对故障的类型和原因进行诊断,以有效对故障进行预防。但是,由于实际情况受多因素影响,不易通过过程的数学模型进行故障诊断。数据驱动是建立在数据分析的基础上,通过运行过程中记录的大量有用的信息进行挖掘,从中获得所需要的信息和知识的过程,具有更加实际的意义。本文研究采用无监督的多元统计分析方法及聚类算法,简单高效地处理工业过程中的故障检测和诊断问题,论文的主要内容包括:首先,分别对核主元分析KPCA和模糊C均值聚类FCM算法进行了详细分析,进一步将KPCA与FCM进行了结合,提出了一类基于KPCA的FCM聚类方法。其次,将所提出的方法应用于工业过程故障诊断问题。核主元分析(KPCA)算法通过映射将数据从输入空间映射到高维特征空间,从而对数据进行特征提取,有效降低数据维度和提高处理效率。同时可通过KPCA结合T2和SPE两个统计量进行实时的故障检测。接着,采用FCM算法对故障模式类别进行划分,从而实现故障的诊断。最后,将提出的这种方法用于TE化工生产过程的仿真,采用KPCA-FCM方法,获得了故障数据的分类,同时实现故障数据的在线诊断。利用该数据驱动算法,可高效检测工业过程中的故障并对故障进行诊断。
[Abstract]:The effective method of industrial process fault diagnosis can not only identify and deal with the fault in the first time, avoid the possible danger, but also diagnose the type and cause of the fault in order to prevent the fault effectively. However, because the actual situation is affected by many factors, it is difficult to carry out fault diagnosis through the mathematical model of the process. Data-driven is a process based on the analysis of data, mining a lot of useful information recorded in the running process, and obtaining the needed information and knowledge from the process, which has more practical significance. In this paper, the unsupervised multivariate statistical analysis method and clustering algorithm are used to deal with the problems of fault detection and diagnosis in industrial process simply and efficiently. The main contents of this paper are as follows: first, The kernel principal component analysis (KPCA) and fuzzy C-means clustering (FCM) algorithms are analyzed in detail. Furthermore, a kind of FCM clustering method based on KPCA is proposed by combining KPCA with FCM. Secondly, the proposed method is applied to industrial process fault diagnosis. The kernel principal component analysis (KPCA) algorithm maps the data from the input space to the high-dimensional feature space, which can effectively reduce the data dimension and improve the processing efficiency. At the same time, real-time fault detection can be carried out by KPCA combined with T2 and SPE statistics. Then, the FCM algorithm is used to divide the fault mode categories to realize fault diagnosis. Finally, the proposed method is applied to the simulation of te chemical production process. The classification of fault data is obtained by using KPCA-FCM method, and the fault data online diagnosis is realized at the same time. Using the data driven algorithm, the fault in industrial process can be detected and diagnosed efficiently.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ086

【相似文献】