-
当复合材料表面及浅表面层存在夹杂、脱粘等内部缺陷时,随着外部负载的变化,缺陷和材料本身性质不同,在缺陷所在的材料外表面会产生变形,使得缺陷处空间位移梯度发生变化。传统数字剪切散斑干涉光路一般为迈克尔逊剪切结构,如图1所示。激光器输出的相干光照射到被测物表面产生漫反射,反射光进入迈克尔逊结构后在倾斜的平面镜2处与入射光产生一个微小的角度错位,最终在图像传感器前发生干涉。文中在迈克尔逊剪切结构前后嵌入焦距相互匹配的透镜1、2组成4f光路,透镜1的前焦面作为输入面,透镜2的后焦面作为输出面,两透镜焦距相匹配,图像大小比例不变,图像能够完整复原[16]。检测视场角为:
$$\alpha = 2{\rm{arcta}}{{\rm{n}}}\frac{h}{{2f}}$$ (1) 式中:
$\alpha $ 为视场角度;h为相机靶镜尺寸;f为成像镜头的焦距,通过选取合适的焦距可以扩大视场角。当激光器和相机被布置在xoz平面,且剪切量在x轴上时,假设入射光与中心轴的夹角很小,根据剪切散斑干涉原理,位移空间梯度可近似表示为:$${\Delta _\varphi }\left( {x,y} \right) \approx \frac{{4\pi \delta x}}{\lambda } \cdot \frac{{\partial w\left( {x,y} \right)}}{{\partial x}}$$ (2) 式中:
${\Delta _\varphi }\left( {x,y} \right)$ 为平面上某一点(x,y)的干涉相位差;$\lambda $ 为激光器的波长;$\delta x$ 为x轴上的剪切量;${{\partial w\left( {x,y} \right)} / {\partial x}}$ 为点(x,y)处z轴位移分量w在x方向上的空间位移梯度。当剪切图像(x,y)处存在图1所示的剪切散斑条纹时,则表明此处位置存在相位差,空间梯度不为零,存在材料缺陷。这就是剪切散斑干涉技术进行缺陷位置和尺寸检测的基本原理。通过公式(2)可以发现,相位差只和空间位移梯度有关,而与位移量无关,因此剪切散斑干涉测量具有对刚体位移不敏感的优点,抗干扰能力强,适合实际现场应用。 -
剪切散斑干涉单次检测范围有限,利用二维移动台进行分视场扫描检测是扩大检测面积的有效手段。然而复合材料表皮多为铝蒙皮和碳纤维板,实物图缺乏能匹配的纹理,见图2(b);剪切干涉图的散斑效应明显,也缺乏能匹配的特征,见图2(c);因此,无法直接利用实物图和散斑图进行多视场间的视场匹配,完成多视场缺陷坐标位置的全局统一。文中利用辅助投影引入额外表面特征,见图2(a),通过分视场间投影图的单应性矩阵计算分视场间实物图和散斑干涉图的全局匹配和坐标统一,形成了投影图—实物图—干涉图的多视场扫描与匹配拼接检测原理。
图 2 图像特征。(a)实物图;(b)散斑干涉图;(c)投影图
Figure 2. Image feature. (a) Physical picture; (b) Speckle pattern interferometry; (c) Projection pattern
对于一个分视场Vi,通过光源切换可以依次采集投影图Ti,实物图Si和干涉图Ii,如图2所示。由于采集过程只切换光源,不改变相机与被测物的相对位置关系,因此如果分视场Vi的投影图Ti和分视场Vj的投影图Tj之间的单应性变换矩阵是Hij,则对应的实物图Si和实物图Sj,以及干涉图Ii和干涉图Ij之间的单应性变换矩阵仍然是Hij,即:
$${T_j} = {H_{ij}}{T_i}$$ (3) $$\begin{split} &\\ {S_j} = {H_{ij}}{S_i}\end{split}$$ (4) $$ {I_j} = {H_{ij}}{I_i} $$ (5) 有共同视场的投影图Ti和Tj利用投射的辅助图案,通过图像特征提取(如Surf特征,Sift特征等[17-18])及其匹配算法可以计算投影图之间的单应性矩阵Hij,再根据公式(4)和(5)建立实物图以及干涉图的多视场匹配与拼接。
-
如图3~4所示,剪切散斑干涉测头、激光光源以及加热灯(加载组件)等三部分作为移动测头被安装在一个二维位移平台上。实验系统中的迈克尔逊剪切装置由两个直径为12.7 mm的反射镜和一个边长为12.7 mm的分光棱镜组成,其中平面镜1后安装一个压电陶瓷作为相移器,4f系统则由两个焦距均为40 mm相互匹配的平凸透镜组成。工业相机采用北京凯视佳光电设备有限公司的MU3HS500M黑白工业相机,相机的靶面尺寸为2/3 inch(1 inch=2.54 cm),成像镜头是日本尼康超广角镜头(尼克尔系列),焦距为14 mm。结合所述器件的参数信息与公式(1)可得,剪切散斑干涉测头在水平方向上的实际视场角可达33.8°,垂直方向上的实际视场角可达28.5°,突破了传统28°的理论值。当工作距离在1.2 m左右时,它能够直接完成被测样件尺寸为600 mm×500 mm面积的缺陷检测。激光器采用的是200 mW单纵模固体激光器(长春新产业MSL系列),中心波长532 nm。在二维移动台视场不受遮挡的支架处固定30 W的投影灯投影设备,投射图案如图5所示,当工作距离为1.5 m左右时投射面积约为2 m2。结合投影辅助及二维运动台通过移动测头系统可实现被测件无损检测的覆盖面积为3.5 m×4.0 m。
-
实验的复合材料由蒙皮厚度0.3 mm 的铝薄板,边长为5 mm的铝蜂窝芯格制作而成,样品尺寸600 mm×500 mm,见图6(a);图4实物系统采集的散斑干涉图像检测结果见图6(b)。
图6(b)所示实物系统检测效果良好,获得的缺陷条纹平滑流畅、背景清晰、缺陷位置和相对大小一目了然。其中缺陷7经图像滤波和解包裹获得的离面位移空间梯度如图7所示,离面位移空间梯度峰峰值为1.36×10−4。通过系统标定,图6(b)中的14处缺陷的位置和尺寸如表1第2列和第5列所示,这些预制缺陷的实际位置和尺寸如表1第3列和第6列所示。结果表明,缺陷定位的均方根误差为7.0 mm,此方法能够实现缺陷相对精确的定位;缺陷尺寸测量的均方根误差为4.9 mm,由于缺陷引起的变形不可能小于缺陷所占面积,因此测量结果会略大于真实值,实际与这一基本事实相符。
表 1 缺陷信息
Table 1. Defect information
Number Defect location/mm True location/mm Location error/mm Defect size/mm True size/mm Size relative error 1 (306.4,93.7) (300.0,100.0) 9.0 33.6 30.0 12.0% 2 (403.1,94.6) (400.0,100.0) 6.2 44.7 40.0 11.8% 3 (505.1,95.8) (500.0,100.0) 6.6 54.9 50.0 9.8% 4 (107.0,198.7) (100.0,200.0) 7.1 52.9 50.0 5.8% 5 (206.4,194.9) (200.0,200.0) 8.2 45.4 40.0 13.5% 6 (305.7,196.8) (300.0,200.0) 6.5 36.7 30.0 22.3% 7 (204.5,297.9) (200.0,300.0) 5.0 28.6 25.0 14.4% 8 (305.7,294.8) (300.0,300.0) 7.7 34.9 30.0 16.3% 9 (406.9,296.9) (400.0,300.0) 7.6 45.3 40.0 13.3% 10 (504.5,298.7) (500.0,300.0) 4.7 56.5 50.0 13.0% 11 (105.1,396.2) (100.0,400.0) 6.4 55.7 50.0 11.4% 12 (206.4,396.9) (200.0,400.0) 7.1 44.8 40.0 12.0% 13 (305.8,397.7) (300.0,400.0) 6.2 33.7 30.0 12.3% 14 (407.0,394.9) (400.0,400.0) 8.6 29.0 25.0 16.0% RMS (5.8,4.0) (0,0) 7.0 4.9 0 13.6% -
采用表面光滑的铝蒙皮-铝蜂窝夹心结构材料作为复合材料实验样品。图8~10的(a)~(c)分别是采集的连续三个视场处的投影图、实物图和散斑干涉图,根据1.2节给出的大尺寸拼接算法,首先图8(a)和(b)通过特征点匹配,可以计算其单应性矩阵为:
图 8 投影图。(a)投影视场1;(b)投影视场2;(c) 投影视场3;(d)投影全场图像
Figure 8. Projection image. (a) Projection field of view 1; (b) Projection field of view 2; (c) Projection field of view 3; (d) Full-field projection image
图 9 实物图。(a)实物视场1;(b) 实物视场2;(c) 实物视场3;(d)实物全场图像
Figure 9. Physical picture. (a) Physical field of view 1; (b) Physical field of view 2; (c) Physical field of view 3; (d) Full-field physical image
图 10 散斑干涉图。(a)干涉图1;(b)干涉图2;(c)干涉图3;(d)全场干涉图像
Figure 10. Speckle pattern interferogram. (a) Interferogram 1; (b) Interferogram 2; (c) Interferogram 3; (d) Full-field interferogram
$$\begin{split} &\\ {H_{{\rm{ab}}}}{\rm{ = }}\left[ {\begin{array}{*{20}{c}} {{\rm{0}}{\rm{.938\;1}}}&{{\rm{ - 0}}{\rm{.025\;3}}}&{{\rm{284}}{\rm{.356\;8}}} \\ {{\rm{ - 0}}{\rm{.030\;8}}}&{{\rm{0}}{\rm{.981\;2}}}&{{\rm{4}}{\rm{.227\;3}}} \\ {{\rm{ - 0}}{\rm{.000\;1}}}&{{\rm{ - 0}}{\rm{.000\;2}}}&{{\rm{1}}{\rm{.000\;0}}} \end{array}} \right] \end{split}$$ 图8(b)和8(c)通过特征点匹配,又得到两视图的单应性矩阵为:
$${H_{{\rm{bc}}}}{\rm{ = }}\left[ {\begin{array}{*{20}{c}} {{\rm{1}}{\rm{.061\;3}}}&{{\rm{ - 0}}{\rm{.009\;9}}}&{{\rm{284}}{\rm{.310\;6}}} \\ {{\rm{0}}{\rm{.018\;1}}}&{{\rm{1}}{\rm{.022\;5}}}&{{\rm{ - 0}}{\rm{.374\;7}}} \\ {{\rm{0}}{\rm{.000\;2}}}&{{\rm{0}}{\rm{.000\;2}}}&{{\rm{1}}{\rm{.000\;0}}} \end{array}} \right]$$ 根据上述计算得到的两个单应性矩阵,投影图进行拼接之后获得拼接后图像(图8(d))。同理,利用上述两个单应性矩阵,根据公式(4)和(5),可以拼接图9~10的(a)~(c)获得实物图和散斑干涉图的拼接后全场图像,见图9(d)~10(d)。
实验采用的复合材料板为实际生产成品,内部并未预制缺陷。其检测结果显示该板被检区域部分无缺陷分布,产品质量良好,检测结果与复合材料板实际情况相符。
另一被测样品为表面粗糙的碳纤维蒙皮-铝蜂窝夹心的圆盘形复合材料,圆盘直径为Φ600 mm。从图11中可以看出,拼接图像完整地还原了整个圆盘的全貌,并且拼接而成的图像没有错误交叉点,实物图仍保留了系统的剪切效果。
图 11 分视场及全场图像。(a)分视场投影图像;(b)分视场实物图像;(c)分视场干涉图像;(e)投影图像拼接;(f)实物图像拼接;(g)干涉图拼接
Figure 11. Split field and full-field images. (a) Sub-field projection image; (b) Sub-field physical image; (c) Sub-field interferogram; (e) Projection image stitching; (f) Physical image stitching; (g) Interferogram stitching
结合上述实验结果,所提出的方法在现有检测技术基础上,扩大了单次视场的检测面积,并通过分视场投影图像完成了单视场下采集的实物图像及散斑干涉图的无缝拼接,有效得到被测复合材料的全尺寸完整图像及对应的检测图像。
Projection aided digital shearography scanning detection technology
-
摘要: 数字剪切散斑干涉技术已被广泛应用于复合材料无损检测,但常规检测面积有限,难以完成大尺寸复合材料的缺陷检测。提出了一种投影辅助数字剪切散斑干涉大尺寸复合材料扫描检测方法,该方法基于数字剪切散斑干涉技术并利用辅助投影引入额外表面特征,通过分视场间投影图的单应性矩阵计算分视场间实物图和散斑干涉图的全局匹配和坐标统一,完成了投影图-实物图-干涉图的多视场扫描与匹配拼接,同时在单视场检测引入4f光路及超广角镜头扩大单次检测面积。实验结果表明:此扫描检测系统能够实现缺陷位置、尺寸较精确的测量,位置定位均方根误差为7.0 mm,尺寸测量误差的均方根值为4.9 mm。在1.2 m的工作距离下单次检测面积可达600 mm×500 mm,全局扫描检测面积高达3.5 m×4.0 m。此方法具备抗刚体位移干扰强,缺陷检测灵敏度高的优点,适合大尺寸高性能复合材料无损检测现场使用。Abstract: Digital shearography has been widely used in non-destructive testing of composite materials. But the conventional detection area is limited and the defect detection of large-size composite materials is difficult to achieve. A digital shearography assisted with projection scanning detection for large-size composite materials was introduced. Based on digital shearography and the auxiliary projection which could project additional surface features, global matching and coordinate unification of the sub-view original images and the speckle interferograms were calculated by the homography matrix of the corresponding projection images to realize the multi-field scanning and matching stitching of projection, origin, and interferogram images. At the same time, a 4f optical path and an ultra-wide-angle lens were imported for the one detection to expand the detection area. Experimental results show that the scanning detection system can achieve more accurate measurement of the defect position and size. The root mean square error of location positioning is 7.0 mm, and the root mean square value of dimension measurement error is 4.9 mm. At a working distance of 1.2 m, the single detection area can reach 600 mm×500 mm and the global scanning detection area can reach 3.5 m×4.0 m. This method has the advantages of strong resistance to rigid body displacement interference and high sensitivity of defect detection, which is suitable for non-destructive testing of large-size and high performance composite materials.
-
图 11 分视场及全场图像。(a)分视场投影图像;(b)分视场实物图像;(c)分视场干涉图像;(e)投影图像拼接;(f)实物图像拼接;(g)干涉图拼接
Figure 11. Split field and full-field images. (a) Sub-field projection image; (b) Sub-field physical image; (c) Sub-field interferogram; (e) Projection image stitching; (f) Physical image stitching; (g) Interferogram stitching
表 1 缺陷信息
Table 1. Defect information
Number Defect location/mm True location/mm Location error/mm Defect size/mm True size/mm Size relative error 1 (306.4,93.7) (300.0,100.0) 9.0 33.6 30.0 12.0% 2 (403.1,94.6) (400.0,100.0) 6.2 44.7 40.0 11.8% 3 (505.1,95.8) (500.0,100.0) 6.6 54.9 50.0 9.8% 4 (107.0,198.7) (100.0,200.0) 7.1 52.9 50.0 5.8% 5 (206.4,194.9) (200.0,200.0) 8.2 45.4 40.0 13.5% 6 (305.7,196.8) (300.0,200.0) 6.5 36.7 30.0 22.3% 7 (204.5,297.9) (200.0,300.0) 5.0 28.6 25.0 14.4% 8 (305.7,294.8) (300.0,300.0) 7.7 34.9 30.0 16.3% 9 (406.9,296.9) (400.0,300.0) 7.6 45.3 40.0 13.3% 10 (504.5,298.7) (500.0,300.0) 4.7 56.5 50.0 13.0% 11 (105.1,396.2) (100.0,400.0) 6.4 55.7 50.0 11.4% 12 (206.4,396.9) (200.0,400.0) 7.1 44.8 40.0 12.0% 13 (305.8,397.7) (300.0,400.0) 6.2 33.7 30.0 12.3% 14 (407.0,394.9) (400.0,400.0) 8.6 29.0 25.0 16.0% RMS (5.8,4.0) (0,0) 7.0 4.9 0 13.6% -
[1] Soutis C, Moreira P. Recent advances in structural integrity of engineering composite materials [J]. Applied Composite Materials, 2020, 27: 447-448. doi: 10.1007/s10443-020-09830-6 [2] Wang B, Zhong S, Lee T L, et al. Non-destructive testing and evaluation of composite materials/structures: A state-of-the-art review [J]. Advances in Mechanical Engineering, 2020, 12(4): 1-28. [3] McIlhagger A, Archer E, McIlhagger R. Polymer Composites in the Aerospace Industry [M]. 2nd ed. Britain: Woodhead Publishing, 2020: 59-81. [4] Harris C E, Starnes J H, Shuart M J. An assessment of the state-of-the-art in the design and manufacturing of large composite structures for aerospace vehicles[C]//Langley Research Center, National Aeronautics and Space Administration, 2001. [5] Diamanti K, Soutis C, Hodgkinson J M. Piezoelectric transducer arrangement for the inspection of large composite structures [J]. Composites Part A: Applied Science & Manufacturing, 2007, 38(4): 1121-1130. [6] El–sabbagh A, Steuernagel L, Ziegmann G. Characterisation of flax polypropylene composites using ultrasonic longitudinal sound wave technique [J]. Composites Part B: Engineering, 2013, 45: 1164-1172. doi: 10.1016/j.compositesb.2012.06.010 [7] Kolkoori S, Wrobel N, Zscherpel U, et al. A new X-ray backscatter imaging technique for non-destructive testing of aerospace materials [J]. Ndt & E International, 2015, 70: 41-52. [8] Kou Guangjie, Yang Zhengwei, Jia Yong, et al. Detection on cracks in blades with complex profile based on ultrasonic infrared thermal imaging [J]. Infrared and Laser Engineering, 2019, 48(12): 1204002. (in Chinese) doi: 10.3788/IRLA201948.1204002 [9] Wang Q, Hu Q P, Qiu J X, et al. Detection of internal defects in aviation composites with differential laser infrared thermal imaging [J]. Infrared and Laser Engineering, 2019, 48(5): 0504003. (in Chinese) doi: 10.3788/IRLA201948.0504003 [10] Towsyfyan H, Biguri A, Boardman R, et al. Successes and challenges in non-destructive testing of aircraft composite structures [J]. Chinese Journal of Aeronautics, 2019, 33(3): 1-30. [11] Ma Y, Jiang H, Dai M, et al. Cantilevered plate vibration analysis based on electronic speckle pattern interferometry and digital shearing speckle pattern interferometry [J]. Acta Optica Sinica, 2019, 39(4): 0403001. doi: 10.3788/AOS201939.0403001 [12] Sun P. Digital phase-shifting shearography for strain measurement by using a rotating platform system [J]. Optical Engineering, 2005, 44(8): 085601. doi: 10.1117/1.2010127 [13] Wang Xin, Wang Yonghong, Lv Youbin, et al. Whole field optical detection method of strain distribution of SU-8 photoresist [J]. Chinese Optics, 2016, 9(3): 379-384. (in Chinese) doi: 10.3788/co.20160903.0379 [14] Wang Yonghong, Lv Youbin, Gao Xinya, et al. Research progress in shearography and its applications [J]. Chinese Optics, 2017, 10(3): 300-309. (in Chinese) doi: 10.3788/co.20171003.0300 [15] Zhao Q, Dan X, Sun F, et al. Digital shearography for NDT: Phase measurement technique and recent developments [J]. Applied Sciences, 2018, 8(12): 2662. doi: 10.3390/app8122662 [16] Wu S, He X, Yang L. Enlarging the angle of view in Michelson-interferometer-based shearography by embedding a 4f system [J]. Applied Optics, 2011, 50(21): 3789-3794. doi: 10.1364/AO.50.003789 [17] Liu Jie, You Pinhong, Zhan Jianbin, et al. Improved SIFT fast image stitching and ghosting optimization algorithm [J]. Optics and Precision Engineering, 2020, 28(9): 2076-2084. (in Chinese) doi: 10.37188/OPE.20202809.2076 [18] Di Hongzhang, Yang Wen, Lin Pengyue, et al. Mosaic of cultural relics fragments based on SURF feature extraction descriptor and Jaccard distance [J]. Optics and Precision Engineering, 2020, 28(4): 963-972. (in Chinese)