基于阶次跟踪和BP网络的齿轮箱故障模式识别

在对齿轮箱的原始振动信号进行时域采样的基础上,利用小波包分析对其进行消噪,然后对得到的数据进行角域重采样,得到基于阶次跟踪的采样信号,再对该信号进行特征参量提取,最后利用BP网络对得到的故障特征参量集进行模式识别。该方法能够避免传统分析方法中难以克服的“频率模糊”现象,对于瞬态信号有较好的分析处理能力,是对传统频谱分析法的有力补充。

Fault Mode Identification of Gearbox Based on Order Tracking and BP Network

Based on time domain sampling on the originality vibration signal of the gearbox,which has been eliminated the confusion of noises by using the wavelet analysis,then the outcomes are resampled in angle domain,at last the sampling data based on the order tracking are received.The fault character parameters are extracted from the signals,and then the character parameters collection are identified with BP network.The methods can avoid effectively the "frequency smear"phenomenon,which cannot be solved with the traditional frequency spectrum method.The result proves that the method has better abililty to analyze and dispose instantaneous signals and the order tracking analysis is a trenchant supplement for traditional spectrum analysis.