模糊数据融合技术在系统故障诊断中的应用

利用系统故障症状的分散性，提出了一种基于模糊数据融合技术的系统故障诊断方法。首先，为保证量测信号的准确性，采用同源多传感器数据层融合以及多传感器信息优化协调技术对量测数据进行初步处理，为系统故障模糊融合诊断奠定基础；然后，针对同一症状的不同表现信息，采用多个模糊神经网络得到对故障的局部决策，利用模糊积分融合方法，识别出该症状所对应的故障。最后针对某液体火箭发动机的泄漏故障进行仿真，并与常规模糊神经网络故障分类器进行对比。研究表明，在对渐变故障的诊断中，本法较常规模糊神经网络故障分类器具有更好的诊断性能。

Application of fuzzy data fusion to fault diagnosis in system

In this paper, we propose a novel approach to fault diagnosis based on fuzzy data fusion technique. First, we use data layer fusion and information optimization technique to confirm the reliability and accuracy of the measured signals. Then, we adopt multiple neuro-fuzzy networks as fault classifiers to obtain the local decisions. Furthermore, we present a method for decision fusion based on fuzzy integral in which the relative importance of the different individual network is consid- ered. Finally, we apply this approach to detecting leak faults in a rocket engine and comparing the results with that of traditional multiple neuro-fuzzy network fau1t classifier.The simulation results illus- trate that it can improve the diagnosis performance evidently.