齿轮损伤诊断的非参数统计检验方法

A Nonparametric Statistics Test Method for Gear Fault Diagnosis

提出一种基于振动信号频谱的双样本Kolmogorov-Smirnov(K-S)检验的齿轮损伤诊断方法。从齿轮箱振动信号中获得齿轮啮合振动的时域同步平均信号及其频谱,与齿轮初始啮合阶段的振动时域同步平均信号及其频谱比较,计算其K-S统计距离和相似概率,并将其作为齿轮磨损的量化指标。针对齿轮全生命周期啮合试验的振动信号数据,计算该量化指标。结果表明:时域信号K-S检验的统计距离可以反映齿轮状态的变化,相似概率的变化无规律,频域的这2个指标值均能指示出齿轮破损。

A Kolmogorov-Smirnov test using gear vibration signal frequency spectrum to detect the occurrence of gear crack is proposed.Time synchronous averaging signal(TSA)and its frequency spectrum of gear vibration were gained from gear box.The K-S test between the TSA frequency spectrum of the detected gear state and the normal gear state was done.The statistics distance and the similarity probability were used as crack feature parameters.For the vibration data of a gear life cycle,the two feature parameters were calculated.The results show that the statistical distance of time domain signal K-S test can reflect the change of gear state,and the change of similarity probability is irregular,and the two index values in frequency domain can indicate the gear damage.