基于AR-TFM方法的超声成像分辨力提升  

Resolution enhancement of ultrasonic imaging based on AR-TFM method

作　　者：史思琪 林莉[1] 严宇 廖静瑜 张晓峰 杨会敏 金士杰[1] SHI Siqi;LIN Li;YAN Yu;LIAO Jingyu;ZHANG Xiaofeng;YANG Huimin;JIN Shijie(NDT&E Laboratory,Dalian University of Technology,Dalian 116024,China;Nuclear Industry Research and Engineering Co.,Ltd.,Beijing 101300,China;China Nuclear Industry 23 Construction Co.,Ltd.,Beijing 101300,China)

机构地区：[1]大连理工大学无损检测研究所,大连116024 [2]核工业工程研究设计有限公司,北京101300 [3]中国核工业二三建设有限公司,北京101300

出　　处：《无损检测》2021年第12期1-4,80,共5页Nondestructive Testing

基　　金：国家自然科学基金(51905079);辽宁省“兴辽英才计划”项目(XLYC1902082)。

摘　　要：将多频带自回归(AR)谱外推和全聚焦方法(TFM)相结合,提出了一种AR-TFM方法,可将超声成像分辨力从波长级提升至亚波长级。采用基于高阶累积量的奇异值分解方法确定AR阶数,选择多组有效频带外推处理全矩阵数据中的阵列信号,并对结果进行平均加权,最后实施延时叠加和逐点成像,提高该方法的适用性和鲁棒性。分别采用TFM和AR-TFM方法对碳钢试块进行检测,并对其检测结果进行对比。仿真和试验结果表明,AR-TFM方法能够突破瑞利准则的限制,有效提高超声成像分辨力,实现了碳钢试块3个中心距为0.7λ(λ为超声波长)的相邻圆孔的分离,缺陷深度定位误差不超过1.8%,中心距定位误差不超过15%。In this paper,the multi-frequency windows autoregressive spectral extrapolation with total focusing method(AR-TFM)is presented and applied to raise the ultrasonic imaging resolution to from wavelength-scale subwavelength-scale.Singular value decomposition based on higher order cumulant statistics is employed to determine the AR order.Full matrix capture data is processed by AR spectrum extrapolation based on several effective frequency bands,and then extrapolated spectra are linearly combined and averaged.The TFM imaging is performed on the extrapolated full matrix capture data to enhance the robustness and applicability.The TFM and AR-TFM methods were used to test the carbon steel test blocks,and the test results were compared.The simulated and experimental results demonstrate that the diffraction limit is broken through by AR-TFM,realizing the identification of three side-drilled holes(SDHs)with 0.7λ(wavelengthλ)central distance.The relative measurement errors of the depths and central distance of SDHs are no more than 1.8%and 15%,respectively.

关 键 词：自回归谱外推 全聚焦方法 成像分辨力 AR-TFM 

分 类 号：TG115.28[金属学及工艺—物理冶金]