Using support vector machines and rough sets theory for classifying faulty types of diesel engine

Abstract

Support vector machines (SVM) and rough sets theory (RST) are two emerging techniques in data analysis. The RST can deal with vague data and remove redundant attributes without losing any information of the data; and SVM has powerful classification ability. In this study, the RST is employed to reduce data attributes. Then, the reduced attributes are used by the SVM model for classification. An example of diesel engine diagnosis in the literature is used to demonstrate the diagnosis ability of the proposed RSSVM (rough set theory with support vector machines) model. In terms of classification accuracy and efficiency, experimental outcomes show that the RSSVM model can provide better diagnosis results than those obtained by the directed acyclic graph support vector machine (DAGSVM) model. © Springer-Verlag Berlin Heidelberg 2007.