Oil Fault Diagnosis Based on Fuzzy Rough Set Theory and Bayesian Network

Abstract

This paper proposes a new method of information fusion of oil fault diagnosis. Firstly a fuzzy decision information system is established using fuzzy processing of oil monitoring data. Aiming at the problem that rough set can not be directly applied to the data with continuous variable, this paper adopts the method of fuzzy information system knowledge discovery to the reduction of attributes, which can avoid the information loss by discretizating continuous attribute values in rough set theory. Then based on the connection between the fault symptoms of diesel and oil monitoring data, this paper constructs a Bayesian diagnosis network with the topological structure being used to express the qualitative knowledge and the probability distributions of the nodes in the network to solve the uncertainty of the knowledge. Finally, an example proves that the great significance that information fusion is used in the field of oil monitoring.