超声相控阵检测是一种重要的无损检测技术，广泛应用于核工业、航空航天、机械制造等领域。随着我国工业制造水平的提高，对相控阵检测能力的要求也越来越高。全矩阵采集（FMC）是一种新的基于相控阵硬件平台的数据采集方法，基于全矩阵采集的全矩阵全聚焦（FMC-TFM）成像算法，具有检测精度高、缺陷表征能力强等特点。然而，FMC-TFM算法忽略了信号中的空间信息，信号的旁瓣抑制较差，图像的分辨率较低。本文研究了基于相位相干的延时组合乘叠加（PC-DMAS）成像算法。

FMC-TFM成像算法基于延时叠加（DAS）波束形成算法，运用费马原理计算像素点与发射阵元和接收阵元的延时飞行时间，进而提取超声信号的幅值累加成像，这一过程只利用了全矩阵信号中的时域信息，图像分辨率较低。本文基于DMAS波束形成算法，推导出全矩阵采集的DMAS成像算法，该算法在波束形成过程中利用信号的空间信息抑制合成波束信号的旁瓣，提高图像分辨率。基于Field II超声仿真工具包、Verasonics超声相控阵平台进行实验，结果表明DMAS算法成像的分辨率、背景噪声抑制均优于FMC-TFM算法。

DMAS成像算法通过引入空间相干，增强了旁瓣抑制效果，提高了成像分辨率，但DMAS成像算法噪声抑制能力较弱，而实验采集的超声信号中包含的结构噪声、系统噪声、环境噪声等都会影响DMAS成像效果。基于超声信号中缺陷回波信号的相位呈一致性分布，噪声信号的相位呈散乱分布，本文运用统计学方法获得FMC信号的相位信息，自适应构建相位相干权重因子，提出了基于相位相干的延时组合乘叠加（PC-DMAS）成像算法。利用Verasonics超声相控阵平台采集了数据并成像，实验结果表明PC-DMAS成像图像的缺陷对比度、图像分辨率与背景噪声抑制上都优于DMAS算法。

虽然PC-DMAS成像效果优于FMC-TFM，但PC-DMAS算法的时间复杂度为，较FMC-TFM的增加了一个数量级，导致PC-DMAS成像效率差，因此，提出基于稀疏矩阵采集的相位相干延时组合乘叠加算法（SMC-PC-DMAS）。通过减少激发阵元数来减少数据量以提高成像效率，以与全矩阵下相同的有效孔径为原则，将探头声场方向图的主瓣宽度与峰值旁瓣比作为约束条件，使用遗传算法求解稀疏阵列的最优排列方式，分析了不同稀疏率下检测图像的空间分辨率、阵列性能指标、对比度等指标，结果表明，SMC-PC-DMAS在仅需PC-DMAS算法50%数据量的条件下，就能够提供与PC-DMAS相似的检测效果。

虽然SMC-PC-DMAS提高了成像效率，但其成像时间仍需近千秒，时效性难以满足需求，为进一步提高成像效率，本文使用CPU+GPU异构设计模型，设计并行算法框架。利用CPU处理复杂指令集的能力，控制算法的流程；利用GPU并行计算能力强的特点，并行加速SMC-PC-DMAS成像算法中各像素点的声压幅值计算。实验结果表明，并行加速后的算法成像时间仅为加速前的0.08%，成像效率显著提升。

Ultrasonic phased array testing is an important non-destructive testing technology widely used in nuclear industry, aerospace, mechanical manufacturing and other fields. With the improvement of our country's industrial manufacturing level, the demand for phased array testing capability is also increasing. Full Matrix Capture (FMC) is a new data acquisition method based on phased array hardware platform, and Full Matrix Capture Total Focus Method (FMC-TFM) is an imaging algorithm based on full matrix data, which has the characteristics of high detection accuracy and strong defect characterization capability. However, the FMC-TFM algorithm ignores the spatial information in the signal, with poor side lobe suppression and low image resolution. This paper studies the Delay Multiply And Sum (DMAS) imaging algorithm based on phase coherence.

The FMC-TFM imaging algorithm is based on the Delay And Sum (DAS) beamforming algorithm, which uses the Fermat's principle to calculate the delayed flight time between pixel points and the transmitting array as well as the receiving elements. Subsequently, the amplitude accumulation of ultrasonic signals is extracted for imaging. This process only uses the time-domain information from the full-matrix signals, resulting in relatively low image resolution. In this paper, we derive the DMAS imaging algorithm based on full matrix capture, leveraging the DMAS beamforming algorithm. This algorithm suppresses the side lobe amplitude of signals by using spatial information from the full-matrix signals during the beamforming process, thereby enhancing image resolution. Experimental validations were conducted using the Field II ultrasound simulation toolkit and the Verasonics ultrasound phased-array platform. The results indicate that the imaging resolution and background noise suppression of the DMAS algorithm are better than the FMC-TFM algorithm.

The DMAS imaging algorithm enhances the sidelobe suppression effect and improves the imaging resolution through the introduction of spatial coherence, but its noise suppression ability is weak. Meanwhile, the structural noise, system noise, and environmental noise included in the ultrasound signal collected by the experiment will all affect the imaging effect of DMAS. Based on the consistent phase distribution of defect echo signals and the scattered phase of noise signals, this paper proposes a phased-coherence-based delaye multiply and sum (PC-DMAS) imaging algorithm by utilizing statistical methods to obtain phase information and constructing adaptive phase coherence weight factors. Experiments using the Verasonics ultrasonic phased array platform demonstrate that PC-DMAS outperforms DMAS in defect contrast, image resolution, and noise suppression.

Although the PC-DMAS imaging algorithm outperforms FMC-TFM in terms of performance, its time complexity is, representing an order of magnitude increase compared to FMC-TFM's , leading to a lower imaging efficiency for PC-DMAS. To address this issue, the Sparse Matrix Collection-based Phase Coherence Delay-and-Sum with Multiplication and Accumulation (SMC-PC-DMAS) algorithm is proposed. This approach aims to enhance imaging efficiency by reducing the amount of data through a decrease in the number of excited elements, while maintaining a consistent effective aperture compared to full-matrix acquisition. To achieve this, the main lobe width and peak side lobe ratio of the probe's sound field pattern are utilized as constraints. A genetic algorithm is employed to determine the optimal arrangement for the sparse array. A comprehensive analysis is conducted to assess various metrics, including spatial resolution, array performance indicators, and contrast of the detected images at different sparsity rates. The results demonstrate that SMC-PC-DMAS can achieve similar detection performance to PC-DMAS while utilizing only 50% of the data required by the PC-DMAS algorithm.

Although SMC-PC-DMAS has improved imaging efficiency, its imaging time still requires nearly a thousand seconds, making it difficult to meet the requirements for timeliness. To further enhance imaging efficiency, this paper employs a CPU+GPU heterogeneous design model to devise a parallel algorithm framework. This framework leverages the CPU's capability in handling complex instruction sets to control the algorithm's flow, while harnessing the GPU's strength in parallel computing to accelerate the calculation of sound pressure amplitudes for each pixel point in the SMC-PC-DMAS imaging algorithm. Experimental results demonstrate that the imaging time of the algorithm after parallel acceleration is only 0.08% of that before acceleration, resulting in a significant improvement in imaging efficiency.

[1] 沙经伟,范孟豹,曹丙花, 等.金属构件硬度的无损检测研究进展与展望[J]. 机械工程学报,2023,59(24):1-17.

[2] 蔡健荣,梁小祥,许骞, 等.采用X射线三维重构技术检测厚皮柑橘的体积可食率[J]. 农业工程学报,2024,40(01):293-300.

[3] 张辉,张邹铨,陈煜嵘, 等.工业铸件缺陷无损检测技术的应用进展与展望[J]. 自动化学报,2022,48(04):935-956.

[4] 田贵云,高斌,高运来, 等.铁路钢轨缺陷伤损巡检与监测技术综述[J]. 仪器仪表学报,2016,37(08):1763-1780.

[5] 张荣繁,倪培君,郭智敏, 等. 复杂型面构件超声检测中机械回程误差的修正方法[J]. 兵工学报,2023,44(8):2432-2440.

[6] 李伟, 邵鑫宇, 张伯莹, 等. 交流电磁场和电磁超声复合无损检测技术研究[J]. 机械工程学报, 2022, 58(16): 153-159.

[7] 章佩佩. 超声无损检测信号处理方法的研究[D]. 安徽: 安徽理工大学, 2018.

[8] 张辉,宋雅男,王耀南, 等.钢轨缺陷无损检测与评估技术综述[J]. 仪器仪表学报,2019,40(02):11-25.

[9] Hwang Y I, Park J, Kim H J, et al. Performance comparison of ultrasonic focusing techniques for phased array ultrasonic inspection of dissimilar metal welds[J]. International Journal of Precision Engineering and Manufacturing, 2019, 20: 525-534.

[10] Mohammadkhani R, Zanotti Fragonara L, Padiyar M J, et al. Improving depth resolution of ultrasonic phased array imaging to inspect aerospace composite structures[J]. Nature reviews Cancer, 2020, 20(2): 559.

[11] Kuo-Hsia Ling,Yiin-Kuen Fuh,Tin-Chun, et al. Effect of welding sequence of a multi-pass temper bead in gas-shielded flux-cored arc welding process: hardness, microstructure, and impact toughness analysis[J]. The International Journal of Advanced Manufacturing Technology,2015,81(5-8).

[12] 黄玉秋. 相控阵超声缺陷检测信号处理研究[D]. 天津: 天津大学, 2014.

[13] 周正干,李洋,周文彬.相控阵超声后处理成像技术研究、应用和发展[J]. 机械工程学报,2016,52(06):1-11.

[14] 陈选民, 文晟. 钢轨焊缝的相控阵超声定点扫查工艺研究[J]. 铁道科学与工程学报, 2018, 15(8): 2148-2154.

[15] 徐天勇, 董晓虎, 李荣超, 等. 复合绝缘子内部缺陷的超声相控阵检测研究[J]. 电力工程技术, 2018, 37(6): 75-79, 114.

[16] 涂春磊, 周晓军, 杨辰龙, 等. 相控阵超声检测DDF优化方法的研究[J]. 控制工程, 2012, 19(3): 535-538.

[17] 王鹏, 王志强, 蔡晖, 等. 基于全聚焦相控阵超声的叶根槽裂纹检测[J]. 振动.测试与诊断, 2020, 40(06): 1172-1177+1236.

[18] 周正干, 彭地, 李洋, 等. 相控阵超声检测技术中的全聚焦成像算法及其校准研究[J]. 机械工程学报, 2015, 51(10): 1-7.

[19] Madhavanunni A N, Panicker M R. Beam multiply and sum: A tradeoff between delay and sum and filtered delay multiply and sum beamforming for ultrafast ultrasound imaging[J]. Biomedical Signal Processing and Control, 2023, 85: 104807.

[20] Luo L, Tan Y, Li J, et al. Wedge two-layer medium ultrasonic plane wave compounding imaging based on sign multiply coherence factor combined with delay multiply and sum beamforming[J]. NDT & E International, 2022, 127: 102601.

[21] 汪认, 赵鹏, 何建英, 等. 基于超声相控阵的耐候钢接头疲劳裂纹动态监测[J]. 焊接学报, 2022, 43(12): 100-104.

[22] 薛利杰, 刘晴岩, 方雨，等. 特种设备在役螺栓超声相控阵检测成像影响因素[J]. 无损检测, 2016, 38(07): 6-10.

[23] 周正干,李文涛,李洋, 等. 相控阵超声水浸C扫描自动检测系统的研制[J]. 机械工程学报,2017,53(12):28-34.

[24] 李冬如. 基于相控阵检测技术的承压小径管焊缝缺陷信号图像重构方法研究[D]. 广州: 华南理工大学, 2018.

[25] 吴晓红. 整体化复合材料结构的超声相控阵检测技术[D]. 杭州: 浙江大学, 2021.

[26] 卢超,劳巾洁,戴翔.带楔块二维面阵列超声相控阵声场特性分析[J]. 声学学报,2014,39(06):714-722.

[27] 宋双官, 于长乐, 李琪, 等. 不同激活孔径和聚焦深度对相控阵超声横波检测灵敏度的影响[J]. 无损检测, 2022, 44(4):3 5-39, 66.

[28] Holmes C, Drinkwater B W, Wilcox P D. The post-processing of ultrasonic array data using the total focusing method[J]. Insight - Non-Destructive Testing and Condition Monitoring, 2004, 46(11): 677–680.

[29] Holmes C, Drinkwater B W, Wilcox P D. Post-processing of the full matrix of ultrasonic transmit-receive array data for non-destructive evaluation[J]. NDT& E International, 2005, 38(8): 701-711.

[30] 王鹏,王志强,蔡晖, 等. 基于全聚焦相控阵超声的叶根槽裂纹检测[J]. 振动、测试与诊断,2020,40(6):1172-1177.

[31] Ponseenivasan S, Kumar A, Rajkumar K V. Anisotropy Corrected FMC/TFM Based Phased Array Ultrasonic Imaging in an Austenitic Buttering Layer[J]. Applied Sciences, 2023, 13(8): 5195.

[32] 潘峥, 张邦杰, 汪洪量, 等. 超声相控阵全聚焦成像技术倾斜裂纹定量模拟[J]. 中国测试, 2020, 46(04): 36-41.

[33] Rabe U, Pudovikov S. Application of the total focusing method for quantitative nondestructive testing of anisotropic welds with ultrasound[J]. tm-Technisches Messen, 2020, 87(6): 438-450.

[34] 关山月,汪小凯,华林, 等.航空机匣环件曲面结构超声全聚焦检测与缺陷定量方法[J]. 中国机械工程,2023,34(08):892-898.

[35] 彭朝勇,胥松柏,杜创洲, 等.航空铝试块疲劳裂纹相控阵超声成像检测[J].北京航空航天大学学报,2022,48(12):2398-2404.

[36] Sednev D A, Soldatov A I, Soldatov A A, et al. The Effect of the Position Determination Error for Flexible Linear Array Elements on the Tomogram Focusing[J]. Sensors, 2023, 23(10): 4757.

[37] 林莉, 史思琪, 孙旭, 等. 基于自回归谱外推的全聚焦成像分辨力提升[J]. 机械工程学报, 2020, 56(22): 8-13+23.

[38] 牛洪涛, 赵天伟, 金士杰, 等. 曲面构件轮廓全聚焦超声成像表征[J]. 无损探伤, 2022, 46(01): 6-9.

[39] 王冲, 毛捷, 廉国选. 阵元固体指向性补偿对超声全聚焦成像的优化研究[J]. 应用声学, 2018, 37(5): 732-737.

[40] 池强强, 胡明慧. 相控阵超声全聚焦成像算法的有限元仿真研究[J]. 声学技术, 2020, 39(02): 176-18

[41] Howells P. Explorations in fixed and adaptive resolution at GE and SURC[J]. IEEE Transactions on Antennas and Propagation, 1976, 24(5): 575-584.

[42] Synnevag J F, Austeng A, Holm S. Benefits of minimum-variance beamforming in medical ultrasound imaging[J]. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 2009, 56(9): 1868-1879.

[43] 王平,许琴,范文政, 等. 超声成像中基于特征空间的前后向最小方差波束形成[J]. 声学学报,2013,38(01):65-70.

[44] Nguyen N Q, Prager R W. A spatial coherence approach to minimum variance beamforming for plane-wave compounding[J]. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 2018, 65(4): 522-534.

[45] 施荣华,孟秋杰,董健, 等. 一种基于对角载入的鲁棒MVDR波束形成算法[J]. 湖南大学学报（自然科学版）,2012,39(9):57-61.

[46] Zhao S, Jones D L, Khoo S, et al. Frequency-domain beamformers using conjugate gradient techniques for speech enhancement[J]. The Journal of the Acoustical Society of America, 2014, 136(3): 1160-1175.

[47] Lim H B, Nhung N T T, Li E P, et al. Confocal microwave imaging for breast cancer detection: Delay-multiply-and-sum image reconstruction algorithm[J]. IEEE Transactions on Biomedical Engineering, 2008, 55(6): 1697-1704

[48] Moubark A M, Alomari Z, Harput S, et al. Enhancement of contrast and resolution of B-mode plane wave imaging (PWI) with non-linear filtered delay multiply and sum (FDMAS) beamforming[C]2016 IEEE International Ultrasonics Symposium (IUS). IEEE, 2016: 1-4.

[49] 王宝宇,刘瑞麟,张淼, 等. 基于虚源的延时乘累加波束形成算法[J]. 东北大学学报(自然科学版),2023,44(04):469-475

[50] 吴斌, 刘萧冰, 焦敬品, 等. 一种基于压缩感知的超声阵列全矩阵数据重构方法[J]. 测控技术, 2021, 40(3): 96-101.

[51] 杨晓霞, 薛彬, 靳世久, 等. 压缩传感在超声相控阵检测系统中的应用研究[J]. 电子测量与仪器学报, 2015, 29(09): 1286-1294.

[52] Xu Q, Wang H, Tian G, et al. Total Focusing Method Approach of Ultrasonic Phased Array Based on Compressed Sensing[J]. Russian Journal of Nondestructive Testing, 2022, 58(5): 355-368.

[53] 贾乐成, 陈世利, 白志亮, 等. 超声全聚焦成像校正模型及加速算法[J]. 仪器仪表学报, 2017, 38(07): 1589-1596.

[54] 胡宏伟,杜剑,李洋, 等. 基于稀疏矩阵的两层介质超声相控阵全聚焦成像[J]. 机械工程学报,2017,53(14):128-135.

[55] Zhu W, Wei Z, Xiang Y, et al. Sparse optimal design of an ultrasonic sensor array for fast TFM based on a discrete slime mold algorithm[J]. IEEE Sensors Journal, 2024.

[56] 曾邱毓晨, 吴文焘, 李平, 等. 渐进式频率-波数域全聚焦超声成像[J].声学学报, 2022, 47(05): 663-674.

[57] 张昊, 陈世利, 贾乐成. 基于超声相控线阵的缺陷全聚焦三维成像[J]. 电子测量与仪器学报, 2016, 30(7): 992-999.

[58] Marmonier M, Robert S, Laurent J, et al. Real-time 3D imaging with Fourier-domain algorithms and matrix arrays applied to non-destructive testing[J]. Ultrasonics, 2022, 124: 106708.

[59] Njiki M, Elouardi A, Bouaziz S, et al. A multi-FPGA architecture-based real-time TFM ultrasound imaging[J]. Journal of Real-Time Image Processing, 2019, 16: 505-521.

[60] 薛文杰. 超声相控阵全聚焦成像实现与降噪方法研究[D]. 江苏:东南大学,2021.

[61] 吴婵娟, 应恺宁 ,戴鹭楠, 等. 基于ZYNQ加速的激光超声全聚焦成像技术研究[J]. 中国激光, 2023, 50(18): 260-270.

[62] Romero-Laorden D, Villazón-Terrazas J, Martinez-Graullera O, et al. Analysis of parallel computing strategies to accelerate ultrasound imaging processes[J]. IEEE Transactions on Parallel and Distributed Systems, 2016, 27(12): 3429-3440.

[63] Lewandowski M, Walczak M, Witek B, et al. A GPU-based portable phased-array system with full-matrix capture[C] 2018 IEEE International Ultrasonics Symposium (IUS). IEEE, 2018: 1-3.

[64] 周熠. 超声相控阵频域全聚焦算法研究[D]. 江苏:东南大学,2022

[65] 彭华. CRH动车轮对超声相控阵全矩阵成像技术研究[D]. 四川:西南交通大学,2014.

[66] Asl B M, Mahloojifar A. Minimum variance beamforming combined with adaptive coherence weighting applied to medical ultrasound imaging[J]. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 2009, 56(9): 1923-1931.

[67] Lediju M A, Trahey G E, Byram B C, et al. Short-lag spatial coherence of backscattered echoes: Imaging characteristics[J]. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 2011, 58(7): 1377-1388.

[68] Matrone G, Savoia A S, Caliano G, et al. The delay multiply and sum beamforming algorithm in ultrasound B-mode medical imaging[J]. IEEE transactions on medical imaging, 2014, 34(4): 940-949.

[69] 张鑫.核电奥氏体不锈钢窄间隙焊缝缺陷超声定性及定量检测[D]. 辽宁:大连理工大学,2020.

[70] Camacho J, Parrilla M, Fritsch C. Phase coherence imaging[J]. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 2009, 56(5): 958-974.

[71] 董明, 李敬, 索永录, 等. 基于改进果蝇优化算法正交匹配追踪的超声信号降噪方法[J]. 电子学报, 2022,50(2): 493-501

[72] Lockwood G R, Li P C, O'Donnell M, et al. Optimizing the radiation pattern of sparse periodic linear arrays[J]. IEEE Transactions on Ultrasonics, Ferroelectrics, and frequency control, 1996, 43(1): 7-14.

[73] 王伟,李春艳,孙韬, 等. 天绘二号01组SAR图像质量评价[J].测绘学报,2022,51(12):2508-2516.