An approach to fault diagnosis based on a hierarchical information fusion scheme

In this paper, a novel approach, based on a hierarchical information fusion scheme and using the different symptoms of the faults in the various locations of a system, to fault diagnosis of the system is present. First, the data fusion of various location sensors in a system is used to guarantee the reliability and accuracy of measurements. Then, the different symptoms of the faults in various locations of a system are classified via multiple neural networks to obtain local decisions. These local decisions are fused by fuzzy integral in which the relative importance of each network is also considered. Finally, we apply this approach to the model of a turbine system. The simulation results verify the effectiveness of the proposed method.