期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2015
期号：MULTICON
页码：899
出版社：S&S Publications
摘要：Cognitive approaches generally rely on machine learning techniques to configure the nominal state andcreate the faulty ones without requiring any a priori information about the fault signature or on fault time profile. a novelcognitive fault diagnosis framework for processes described by nonlinear dynamic systems that inspects changes in theexisting relationships among sensors. The proposed framework is based on an evolving clustering algorithm that operatesin the parameter space of time invariant linear models approximating such relationships. FDS, which extends the solutionpresented in, relies on a novel evolving-clustering algorithm (ECA) able to learn the nominal state of the process during aninitial training phase create, update, and maintain the fault. A distributed localized faulty sensor detection algorithmtechnology used for evaluation of the proposed solutions reveals the node fault as well as the significant benefits obtainedfrom the sensor networks.
