关键词: 决策级融合；故障诊断；D-S证据理论；BP神经网络；综合关联度

Abstract:

Directing to the low precision of single fault diagnosis method,a decision-level information fusion diagnosis system model was proposed,in which,the method based on the BP neural network and the method based on synthetically relational analysis were applied to the fault diagnosis firstly,and then the Dempster-Shafer evidence theory was utilized to partial fusions in diagnostic results,which were acquired in different test-points at the same time.At last,the decision fusion was applied to different partial fusions and got the finally diagnostic result.The sauction fan fault diagnosis example shows the validity of the dicision-level fusion fault diagnosis model.

Key words: decision-level information fusion;fault diagnosis;D-S evidence theory;synthetically relational analysis