TOFD图像中焊缝埋藏缺陷的智能评定与分级  

Intelligent Evaluation and Classification of Embedded Weld Flaws in TOFD Images

作　　者：余焕伟 任绪凯 廖晓平 欧阳星峰 杜锡勇 Yu Huanwei;Ren Xukai;Liao Xiaoping;Ouyang Xingfeng;Du Xiyong(Shaoxing Special Equipment Testing Institute, Shaoxing 312071;Shaoxing Key Laboratory of Special Equipment Intelligent Testing and Evaluation, Shaoxing 312071;Zhejiang Deli Equipment CO,Ltd.,Shaoxing 312599)

机构地区：[1]绍兴市特种设备检测院,绍兴312071 [2]绍兴市特种设备智能检测与评价重点实验室,绍兴312071 [3]浙江德力装备有限公司,绍兴312599

出　　处：《中国特种设备安全》2024年第3期88-94,共7页China Special Equipment Safety

基　　金：浙江省市场监督管理局科技计划项目“TOFD图像辅助分析及缺陷智能识别系统的研究与开发”(20200333);浙江省市场监督管理局雏鹰计划培育项目“基于正演方法的承压设备缺陷定量表征与评价技术研究”(CY2023215);中国博士后科学基金(2023M742598)。

摘　　要：TOFD检测中缺陷的评定与分级主要依靠人工进行,需要消耗大量的精力和时间,而且还存在误判的可能性。针对该问题,本文提出一套不依赖人工干预的缺陷自动评定与质量分级方法,包括TOFD图像分区与标定、灵敏度评价、缺陷区域分割、评定与分级。试验结果表明,本文方法与传统人工方法相比,全程无须人工干预,能显著提高分析评定人员的工作效率,特别适合大型特种设备制造及安装过程中批量化TOFD图像的缺陷评定与质量分级。In TOFD detection,the quantitative evaluation and classification of flaws mainly carried out manually,consumes a lot of energy and time,and there is the possibility of misjudgment.Aiming at the above problem,a set of automatic defect assessment and quality classification methods which didn’t depend on human intervention were proposed,including TOFD image partition,automatic calibration,image sensitivity evaluation,defect region segmentation,evaluation and quality classification.The experimental results showed that the proposed methods didn’t need human intervention in the whole process compared to the traditional methods and could significantly improve the processing efficiency of batch TOFD images during the manufacture and installation of large special equipment.

关 键 词：衍射时差超声检测 焊缝埋藏缺陷 评定与分级 阈值分割 

分 类 号：X924.2[环境科学与工程—安全科学]