基于D-S推理的汽轮发电机组融合诊断

Fault Diagnosis Based on the Dempster-Shafer Theory of Evidence for Turbine Generator Set

依据汽轮发电机组的故障特性,提出了一种有效的融合算法。首先将多个传感器获得的振动信号进行特征提取,而后通过BP神经网络实现故障分类,最后根据D-S证据推理做出故障决策并给出实例。文中对融合中每个传感器的权重也进行了讨论。

According the fault attribute of rotating machines, this paper presents a effective fusion measure. Firstly, We extract features of vibration-signal offered by different sensors respectively, then classify various fault using BP (Back-Propagation) neural networks. Finally the outputs of BP network of all the sensors are combined through Dempster-Shafer theory of evidence. In this paper, weight of sensor also is argued.