飞机结构X射线裂纹图像智能评定

Intelligent evaluation of X-ray crack image of aircraft structure

飞机结构X射线图像评定过程存在复杂背景下裂纹分割不准确、检出难等问题。基于高效层聚合网络提出一种飞机结构X射线裂纹图像智能评定模型(ELAN-Seg),将ELAN-Seg模型和DeepLabv3+模型的射线图像裂纹分割能力进行对比,结合图像处理技术对模型分割的裂纹长度进行评估,利用飞机强度试验及外场维护过程采集的X射线图像对模型进行验证。结果表明:分割的最小裂纹长度约为3 mm,ELAN-Seg模型对复杂背景射线图像裂纹分割更加准确,裂纹漏检率小于3.8%,该模型具有工程适用性。

In the process of aircraft structure X-ray images evaluation,there are some problems such as inaccurate crack segmentation and difficult crack detection under complex background.An intelligent evaluation model(ELAN-Seg)for X-ray crack images of aircraft structures is proposed based on efficient layer aggregation network.The ELAN-Seg model is compared with the DeepLabv3+model in the crack segmentation ability of the Xray images.Combined with image processing technology,the crack length of model segmentation is evaluated,and the model is verified by X-ray images acquired during aircraft strength test and field maintenance.The results show that the segmented minimum crack length is about 3 mm,the ELAN-Seg model is more accurate in crack segmentation of complex background X-ray images,and the crack detection rate is less than 3.8%.The proposed model has engineering applicability.