Page 62 - 无损检测2024年第十一期
P. 62

2024
                    远东无损检测新技术论坛论文精选
              2024 远东无损检测新技术论坛论文精选
              DOI:10.11973/wsjc240387


                   基于 SRGAN 的介电材料缺陷微波检测图像


                                                     稀疏重建




                                      危洪波 ,李 勇 ,王若男 ,寇 威 ,方 阳 ,陈振茂                    1
                                                                              1
                                                     1
                                            1
                                                             1
                                                                     2
              (1. 复杂服役环境重大装备结构强度与寿命全国重点试验室,陕西省无损检测与结构完整性评价工程技术研究
                   中心,航天航空学院,西安交通大学,西安 710049;2. 陕西省特种设备检验检测研究院,西安 710048)
                       摘  要:介电材料在制造与服役过程中,可能由于制造瑕疵或复杂的服役环境,会出现脱黏、分层、
                   材料损失等缺陷。微波无损检测是评估介电材料结构完整性的有效手段,但高分辨率成像需求与数据
                   处理难度、检测效率之间的矛盾亟待解决。图像稀疏重建算法为解决此问题提供了可能,其中SRGAN
                  (Super-resolution generative adversarial network)模型在稀疏重建方面表现优异。采用SRGAN进行微
                   波成像稀疏重建,并针对微波图像的特性对SRGAN网络进行了改进,通过对比分析验证了改进后的算
                   法在介电材料缺陷稀疏重建中的有效性。试验结果表明,改进后的SRGAN能够显著提高微波图像
                   成像质量,准确还原缺陷细节,为介电材料结构的安全评估和使用寿命预测提供了有力支持。
                       关键词:电磁无损检测;微波检测;介电材料;缺陷成像;深度学习
                       中图分类号:TG115.28      文献标志码:A    文章编号:1000-6656(2024)11-0028-08

               SRGAN-based sparse reconstruction of defects in dielectric materials for microwave detection


                             WEI Hongbo , LI Yong , WANG Ruonan , KOU Wei , FANG Yang , CHEN Zhenmao 1
                                                             1
                                                                      2
                                                                                1
                                       1
                                               1
              (1. State Key Laboratory for Strength and Vibration of Mechanical Structures, Shaanxi Engineering Research Centre of NDT and
                   Structural Integrity Evaluation, School of Aerospace Engineering, Xi’an Jiaotong University, Xi’an 710049,  China;
                              2. Shaanxi Special Equipment Inspection and Testing Institute, Xi’an 710048, China)
                      Abstract: Dielectric materials may occur during the manufacturing and service process due to manufacturing defects or
                   complex service environments, which seriously affects the performance, microwave nondestructive testing is an effective means
                   to assess the structural integrity of dielectric materials, but the contradiction between the demand for high-resolution imaging
                   and the difficulty of data processing and detection efficiency needs to be resolved. Image sparse reconstruction algorithms
                   provide a possibility to solve this problem, among which SRGAN (Super-Resolution Generative Adversarial Network) model
                   is excellent in sparse reconstruction. In this paper, SRGAN was used for microwave imaging sparse reconstruction, and the
                   SRGAN network was improved for the characteristics of microwave images, and the effectiveness of the improved algorithm
                   in the sparse reconstruction of defects in dielectric materials was verified through comparative analysis. The results showed
                   that the improved SRGAN can significantly improved the microwave image quality and accurately restore the defect details,
                   which provides strong support for the safety assessment and service life prediction of dielectric material structures.
                                                                Key words: electromagnetic nondestructive testing; microwave
                                                                inspection; dielectric material; defect imaging; deep learning
                 收稿日期:2024-08-16

                 基金项目:国家科技部磁约束核聚变能发展研究专项项目
             (2019YFE03130003) ;国家自然科学基金项目(52177007,52311540018,     介电材料如玻璃纤维增强聚合物(Glass fiber
              52107009,11927801)                                reinforced polymer,GFRP)、聚乙烯(Polyethylene,
                 作者简介:危洪波(2001—),男,硕士研究生,专业方向为微波无
                                                                PE)具有轻质、强度高、抗腐蚀性能好等特点,被广
              损检测,weihb@stu.xjtu.edu.cn                                                               [1-2]
                 通信作者:李 勇(1978—),男,教授,专业方向为机械结构无损               泛应用于工业制造、能源、交通运输领域                      。但介
                                                                电材料在冲击、腐蚀或者制造缺陷的影响下,可能
              定量检测及完整性评价,yong.li@mail.xjtu.edu.cn
                28
                     2024 年 第 46 卷 第 11 期
                     无损检测
   57   58   59   60   61   62   63   64   65   66   67