A novel fault diagnosis method for boiler drum water level based on rough sets and evidence theory
Qingzhong Gao, Changyong Yin, Guanliang Dong · 2012
As is well-known, there are a lot of uncertainty and incomplete information in the boiler drum water level control system, which brings many troubles to realize the fault diagnosis effectively. Based on the drum water level sensor signals, combining rough sets theory, D-S evidence theory and data fusion technology, this paper proposes a novel fault diagnosis method for the boiler drum water level using BP neural networks. Utilizing the strong fault tolerance of rough set, the drum level sensor signals are considered as a set of condition attributes of fault classification and some reduction decision table based on BP neural networks. The diagnostic capabilities, the diagnostic accuracy and reliability are improved apparently by the formation of multiple independent diagnostic network and evidence theory of information fusion, which takes advantage of redundant information better.