基于D-S证据理论和BP神经网络的齿轮箱故障诊断

Fault Diagnosis for Gearbox Based on D-S Evidential Theory and BP Neutral Networks

在故障诊断时,需要从多方面获得关于同一对象的多维信息并进行融合,才能对设备进行更可靠更准确地诊断,以求得最佳诊断结果。以齿轮箱故障作为研究对象,提出了一种基于D-S证据理论和BP神经网络相结合的信息融合诊断方法,并进行了验证。首先利用BP神经网络对测量数据进行分析诊断,最后用D-S理论对诊断结果进行融合,结果满足需求,从而证明了D-S理论和BP神经网络相结合的诊断方法的实效性。

When Aimedat the fault diagnosis, we need to research the multi-dimensional information about the same object from many aspects, integration, and can diagnose more reliable and more accurately, achieving optimal diagnosis. We regard gearbox as research subjects, presented one diagnosis method based on DS evidential theory and the BP neural network, and verified. Firstly using BP neural network analysis and diagnose measurement data, finally fusion of diagnostic results, the results proved to meet the demand, thus proving the effectiveness of the integrated diagnostic method.