基于分级概率成像的大电机定子绝缘损伤识别

为了正确评估大电机定子绝缘老化程度,提出了一种基于导波复合特征的分级概率成像损伤检测方法。通过提取损伤前后Lamb波信号之间的相关系数,利用全局概率成像初步获取绝缘损伤的分布区域和损伤程度。通过提取Lamb波散射信号波包传播时间和峰值特征,采用局部概率成像方法进一步表征损伤的局部特征。通过对两种损伤概率成像结果进行图像融合获得定子绝缘损伤识别结果。最后,对不同的典型绝缘损伤进行了损伤检测实验。结果表明:利用复合特征和分级概率成像方法可以识别出定子绝缘损伤位置和损伤程度,能够为大电机定子绝缘故障诊断提供更加有效的参考信息。

Stator insulation damage identification in large generators based on a hierarchical probability imaging method

In order to correctly evaluate the aging degree of the stator insulation of large generator, a hierarchical probabilistic imaging damage detection algorithm based on the hybrid features of Lamb wave is proposed in this paper. The correlation coefficients between healthy and damaged Lamb wave signals are extracted and the global probability imaging is adopted to preliminarily estimate the region and degree of the insulation damage. The time of flight and peak value of the scattered Lamb wave packet are extracted and the local probability imaging is adopted to further determine the local feature of the damage. The image fusion of two damage probability imaging results is conducted, and the stator insulation damage identification result is obtained to aggregate and strengthen the damage features. Moreover, the damage detection experiments were conducted to detect different typical stator insulation damages. The results show that the proposed method is capable of identifying both the location and severity of the insulation damage, which can provide effective and valuable information for the fault diagnosis of large generator stator insulation.