A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates

Abstract

The present invention discloses a nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates. The current nondestructive testing methods for composite materials have low testing efficiency and high testing costs. The present invention uses a camera to shoot a laminate under the irradiation of a light source, uses an adaptive threshold binarization algorithm to separate the damaged area and the undamaged area of ​​the entire laminate, and uses image fusion combined with a genetic algorithm to process the edge of the damaged area to obtain an accurate overall damaged area of ​​the laminate; uses the OpenCV database to process each imaging photograph obtained under different light source brightness, separates the complete image of the damaged area of ​​each layer of the laminate, and then uses a hyperbolic tangent function to fit the edge portion of each group of complete images of the damaged area of ​​the same layer to obtain a high-definition image of the damaged area of ​​each layer of the laminate, and finally superimposes the high-definition images of the damaged area of ​​each layer of the laminate to obtain a 3D damage model of the laminate.

Description

Technical Field

[0001] The invention belongs to the field of non-destructive testing, and in particular relates to a non-destructive testing method based on the light transmittance characteristics of a thermoplastic material laminate. Background Art

[0002] Composite materials are made up of two or more different materials. Laminates are one type of composite materials. Due to their light weight, low density and excellent mechanical properties, they are widely used in aerospace, military, manufacturing and other fields. However, aircraft fuselages and wings made of laminates are often damaged by the impact of hail or raindrops, and this situation cannot be avoided. At the same time, since composite materials are anisotropic materials, unlike isotropic materials, their damage mechanism is difficult to study clearly. Therefore, damage detection and prediction can only rely on some physical means. The damage of laminates generally has no obvious characteristics in appearance, and there may be damage such as matrix delamination and fiber breakage inside, which requires detection through non-destructive testing technology.

[0003] Nondestructive testing technology mainly refers to obtaining the internal structure and property changes of damaged objects through media such as light, heat, electricity, magnetism, and sound without changing the structure and properties of the object. The main method of detection through heat is thermal imaging technology, which has high requirements for the heat released by the heat source; the main technology for detection through electricity is eddy current technology, but it can only detect conductive objects and has a small scope of application; the method of detection through magnetism mainly uses the characteristics of magnetism to obtain the main characteristics of damage according to the stress differences of cracks and other defects. This detection method has a simple structure and is suitable for damage detection of large objects, but its accuracy is low and it cannot be used for some precision detection; the method of non-destructive detection through sound is mainly ultrasonic detection, which uses the difference in the acoustic properties of the material to reflect the ultrasonic propagation wave pattern and the difference in propagation speed to obtain the internal defects of the material; the main non-destructive detection technology through optics is laser holographic detection, which uses the interference phenomenon of light to compare the two light wave arrays before and after deformation to obtain damage information. This technology has a complex structure and high detection accuracy, but it has high requirements for operators; there is also detection through computer neural network technology, which collects a large number of damage images, then processes these data through machine learning, and then predicts the damage situation based on big data to obtain results, but the accuracy of the results depends on the collected big data, and it is still in the research stage.

[0004] At present, the main methods for non-destructive testing of composite materials are detection methods based on thermal imaging technology and detection methods based on ultrasound, but the detection efficiency is low and the detection cost is high. Therefore, a new composite material damage detection method with simple structure, low cost and high detection efficiency is needed. Summary of the invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention comprises the following steps:

[0008] Step 1: Wipe the surface of the laminate, and then dry the moisture on the surface of the laminate; then fix the laminate to the laminate fixing plate, with the side of the laminate with the smaller damaged area facing downward.

[0009] Step 2: Fix the camera on the camera stand and place it above the laminate. Turn on the light source at the bottom of the laminate fixing plate and cover the light source with a light shield. The light from the light source passes through the light-transmitting port at the top of the light shield and the light-transmitting slot of the laminate fixing plate and shines on the bottom of the laminate. Observe the imaging effect of the camera through the computer, adjust the position of the camera stand on the camera fixing plate, align the camera's shooting center with the center of the damaged area, and then fix the camera stand.

[0010] Step 3: gradually increase the brightness of the light source and observe the imaging effect of the camera through the computer until there is a light intensity difference between the boundary of the damaged area of ​​the laminate and other parts in the imaging photo transmitted by the camera to the computer. Save the light source brightness at this time as the initial light source brightness in the computer, and also save the imaging photo at this time in the computer.

[0011] Step 4: Increase the brightness of the light source to the preset value, let it stand for a preset time, and save the image in the computer.

[0012] Step 5: Repeat step 4 multiple times to obtain multiple imaging photos under different light source brightness; then, the computer calculates the brightness of all pixels in each imaging photo.

[0013] Step 6: The computer uses an adaptive threshold binarization algorithm to process each imaging photo, so that the damaged area of ​​the layer with the largest damaged area of ​​the laminate in each imaging photo is processed into a bright part, and the rest of the laminate is processed into a dark part.

[0014] Step 7: Extract the damaged area image of each imaging photo under different light source brightness.

[0015] Step 8: The damaged area images obtained in step 7 are fused to obtain a superimposed damaged area image. Then, the edge of the superimposed damaged area image is fitted using a genetic algorithm. Finally, the edge of the superimposed damaged area image after fitting is superimposed with the bright part of the superimposed damaged area image to obtain the final laminate overall damaged area image.

[0016] Step nine: Use the OpenCV database to convert the imaging photos obtained in step five under different light source brightness into digital data, and then set a brightness separation range for the blocked parts of the damaged areas of each layer of the laminate from bottom to top, and separate the pixel data of the damaged area of ​​each layer of each imaging photo of the laminate that is larger than the damaged area of ​​the lower layer. Except for the bottom layer, the data of the damaged area of ​​each other layer is the pixel data of the damaged area of ​​the layer that is larger than the damaged area of ​​the lower layer, the pixel data of the damaged area of ​​each layer below the layer that is larger than the damaged area of ​​the lower layer, and the pixel data of the damaged area of ​​the bottom layer. The damaged area of ​​the bottom layer is the pixel with the highest brightness in the imaging photo; the data of the damaged area of ​​each layer of the laminate is converted into an image to obtain a complete image of the damaged area of ​​each layer of the laminate in each imaging photo.

[0017] Step 10: For the complete images of the damaged areas of each layer obtained in step 9, the complete images of the damaged areas of each group located in the same layer of the laminate are fitted using a hyperbolic tangent function fitting algorithm, and finally a high-definition image of the damaged area of ​​each layer of the laminate is obtained.

[0018] Step 11: The high-definition images of the damaged area of ​​each layer of the laminate obtained in step 10 are superimposed together, and the high-definition images of the damaged area of ​​each layer are represented by different colors, and finally a 3D damage model of the laminate is obtained.

[0019] Preferably, the color of the light source is red.

[0020] More preferably, the maximum brightness of the light source is 3000 nit, and the luminous flux is set to 1200 LM.

[0021] Preferably, in step seven, the damaged area of ​​the imaging photo is extracted, and the specific process is: use the image processing software Photoshop, click the lock polygon tool, use the mouse to select a lock so that the lock includes the bright part and the light and dark boundary part in the image, then select the undamaged area through the "Select Inverse" command, click Delete, and save the separated damaged area image.

[0022] Preferably, the process of fitting the edge of the superimposed damaged area image using a genetic algorithm in step eight is as follows: ① dividing the edge of the superimposed damaged area image into edge regions to obtain an initial population; ② calculating a fitness value f(T)=w1(T)*w2(T)*[u1(T)-u2(T)] for each edge region. 2, where T represents the gray value threshold, w1(T) represents the number of pixels in the edge area whose gray values ​​are less than the gray value threshold, w2(T) represents the number of pixels in the edge area whose gray values ​​are greater than or equal to the gray value threshold, u1(T) represents the average gray value of all pixels in the edge area whose gray values ​​are less than the gray value threshold, and u2(T) represents the average gray value of all pixels in the edge area whose gray values ​​are greater than or equal to the gray value threshold; ③ Determine whether the fitness value of each edge area reaches the average fitness value The sum of the fitness values ​​of all edge regions divided by the total number of edge regions, select the edge regions whose fitness values ​​reach the average fitness value as the next generation population; ④ Calculate the average fitness value of the new generation population Select the edge area whose fitness value in the new generation population reaches the average fitness value as the next generation population; ⑤ Repeat step ④ for more than three times to obtain the edge of the superimposed damaged area image.

[0023] The beneficial effects of the present invention are as follows:

[0024] The present invention proposes a new nondestructive testing method based on the light transmittance of the laminate. The laminate is photographed by a camera under the illumination of a light source, and then an adaptive threshold binarization algorithm is used to separate the damaged area and the undamaged area of ​​the laminate as a whole. The image fusion technology is combined with a genetic algorithm to process the edge of the damaged area, and finally a relatively accurate overall damaged area of ​​the laminate is obtained, thereby studying the relationship between the damaged area of ​​the laminate and the impact force. Most of the pixels in the undamaged area are manually removed before image fusion, which greatly reduces the calculation amount of image fusion and genetic algorithm. Further, the present invention uses the OpenCV database to process each imaging photo obtained under different light source brightness, separates the complete image of the damaged area of ​​each layer of the laminate, and then uses the hyperbolic tangent function to fit the edge part of each group of complete images of the damaged area of ​​the same layer to obtain a high-definition image of the damaged area of ​​each layer of the laminate. Finally, the 3D drawing software Material Studio is used to superimpose the high-definition images of the damaged area of ​​each layer of the laminate to obtain a 3D damage model of the laminate. The model can be used to study the damage trend of each layer of the laminate, and the material of each layer can be improved in a targeted manner to improve the overall performance of the laminate. Compared with existing detection methods, the present invention has a simple system structure, low cost, high detection efficiency and fast result generation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a schematic diagram of the structure of the detection system used in the present invention after the light shield is removed;

[0026] Figure 2 It is a schematic diagram of the overall structure of the detection system used in the present invention;

[0027] Figure 3 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0028] The present invention is further described below in conjunction with the accompanying drawings.

[0029] like Figure 1 and Figure 2 As shown, the detection system used in the non-destructive testing method based on the light transmittance characteristics of thermoplastic laminates of the present invention includes a fixing frame, a camera fixing plate 1, a camera 2, a laminate fixing plate 4, a light source 6 and a light shield 3; the camera fixing plate 1 and the laminate fixing plate 4 are both horizontally arranged, and the upper and lower spacings are fixed on one side of the fixing frame; the camera 2 is fixed to the bottom of the camera fixing plate 1 through the camera frame; the camera frame and the camera fixing plate 1 form a sliding pair, and are fixedly connected by bolts and nuts; the laminate fixing plate 4 is provided with a light-transmitting groove; the light source 6 is located at the bottom of the laminate fixing plate 4 and is fixed to the fixing frame; the light source 6 is provided with a light adjustment knob 5 for adjusting the switch and brightness of the light source; the light shield 3 covers the light source 6, and a light-transmitting port is provided on the top of the light shield 3; the camera 2 adopts an industrial camera that is slightly sensitive to light, and can adopt the camera model MV-CA013-20GC-GIGE of Hangzhou Hikvision Digital Technology Co., Ltd.; the light source 6 adopts a parallel light source with adjustable brightness.

[0030] like Figure 3 As shown, the nondestructive testing method based on the light transmittance characteristics of the thermoplastic material laminate comprises the following steps:

[0031] Step 1: Wipe the surface of the laminate clean to ensure that there is no dust, and then use a dryer to dry the moisture on the surface of the laminate to increase the light transmittance of the laminate and reduce errors; then fix the laminate to the laminate fixing plate 4, with the side of the laminate with a smaller damaged area facing downward.

[0032] Step 2: Fix the camera 2 on the camera stand and place it above the laminate. Turn on the light source 6 at the bottom of the laminate fixing plate 4 and cover the light source 6 with the light shield 3. The light from the light source 6 passes through the light-transmitting port at the top of the light shield 3 and the light-transmitting slot of the laminate fixing plate 4 and irradiates the bottom of the laminate. Observe the imaging effect of the camera 2 through the computer, adjust the position of the camera stand on the camera fixing plate 1, make the damaged area of ​​the laminate within the imaging range of the camera 2, and make the shooting center of the camera 2 as close to the center of the damaged area as possible, and then fix the camera stand.

[0033] Step three: gradually increase the brightness of light source 6, and observe the imaging effect of camera 2 through the computer to ensure that there is enough light to pass through the boundary of the damaged area of ​​the laminate, until there is an obvious light intensity difference between the boundary of the damaged area of ​​the laminate and other parts in the imaging photo transmitted to the computer by camera 2, and then save the brightness of light source 6 at this time in the computer as the initial light source brightness, and also save the imaging photo at this time in the computer.

[0034] Step 4: Increase the brightness of the light source to a preset value (e.g. 200 nit), let it sit for a preset time (usually one minute), and then save the image to the computer.

[0035] Step 5: Repeat step 4 multiple times to obtain multiple imaging photos under different light source brightness; then, the computer calculates the brightness of all pixels in each imaging photo. The brightness calculation is specifically to convert the RGB color space into the YIQ color space, where Y is brightness, Y=0.299R+0.587G+0.114B.

[0036] Step 6: The computer uses an adaptive threshold binarization algorithm (specifically the Wellner algorithm is used in this embodiment) to process each imaging photo, so that the damaged area of ​​the layer with the largest damaged area of ​​the laminate in each imaging photo is processed into a bright part, and the rest of the laminate is processed into a dark part.

[0037] Step 7: Extract the damaged area image of each imaging photo under different light source brightness; extract the damaged area of ​​the imaging photo, the specific process can be: use the image processing software Photoshop, click the lock polygon tool, use the mouse to select a lock, so that the lock includes the bright part and the light and dark boundary part in the image, and then select the undamaged area through the "Select Inverse" command, click Delete, and save the separated damaged area image.

[0038] Step 8: Superimpose the damaged area images obtained in step 7 through the "Superposition" command in the image processing software Photoshop, so that the same parts of the damaged area images are merged and the different parts are retained to obtain the superimposed damaged area image. Since the damaged area images of each imaging photo selected in step 7 will contain some undamaged areas, the superimposed damaged area image after superposition also contains some undamaged areas, and the real edge of the superimposed damaged area image is the area where a large number of bright and dark pixels are distributed alternately. The undamaged area needs to be excluded. The edge can be fitted by setting the gray value threshold. Therefore, the genetic algorithm is used to fit the edge of the superimposed damaged area image, as follows: ① Divide the edge of the superimposed damaged area image into edge areas to obtain the initial population; ② Calculate the fitness value f(T) = w1(T)*w2(T)*[u1(T)-u2(T)] for each edge area 2, where T represents the gray value threshold (some edge points of the real edge can be manually selected, and the gray value average of these edge points is used as the gray value threshold), w1(T) represents the number of pixels in the edge area whose gray value is less than the gray value threshold, w2(T) represents the number of pixels in the edge area whose gray value is greater than or equal to the gray value threshold, u1(T) represents the average gray value of all pixels in the edge area whose gray value is less than the gray value threshold, and u2(T) represents the average gray value of all pixels in the edge area whose gray value is greater than or equal to the gray value threshold; ③ Determine whether the fitness value of each edge area reaches the average fitness value The sum of the fitness values ​​of all edge regions divided by the total number of edge regions, select the edge regions whose fitness values ​​reach the average fitness value as the next generation population; ④ Calculate the average fitness value of the new generation population Select the edge area whose fitness value in the new generation population reaches the average fitness value as the next generation population; ⑤ Repeat step ④ for more than three times to obtain the edge of the superimposed damage area image. Finally, superimpose the edge of the fitted superimposed damage area image with the bright part of the superimposed damage area image to obtain the final laminate overall damage area image.

[0039] Step nine: Use the OpenCV database to convert the imaging photos obtained in step five under different light source brightness into digital data. Then, according to the characteristics that the pixel brightness of the bottom layer of the laminate in the imaging photo is the largest and the pixel brightness of the damaged area blocked by the undamaged area of ​​the lower layer is smaller, a brightness separation range is set for the blocked part of the damaged area of ​​each layer from bottom to top of the laminate, and the pixel data of the damaged area of ​​each layer larger than the damaged area of ​​the lower layer in each imaging photo of the laminate are separated. Except for the bottom layer, the data of the damaged area of ​​each layer is the superposition of the pixel data of the damaged area of ​​the layer larger than the damaged area of ​​the lower layer, the pixel data of the damaged area of ​​each layer below the layer larger than the damaged area of ​​the lower layer, and the pixel data of the damaged area of ​​the bottom layer. The damaged area of ​​the bottom layer is the pixel with the largest brightness in the imaging photo. The data of the damaged area of ​​each layer of the laminate are converted into an image to obtain a complete image of the damaged area of ​​each layer of the laminate in each imaging photo.

[0040] Step 10: For the complete images of the damaged areas of each layer obtained in step 9, the complete images of each group of damaged areas located on the same layer of the laminate are fitted using a hyperbolic tangent function fitting algorithm (which is automatically fitted based on the brightness value of the pixel points), and finally a high-definition image of the damaged area of ​​each layer of the laminate is obtained. Among them, the greater the brightness of the light source, the clearest position in the imaging photo of camera 2 is the position of the layer of the laminate that is farther away from the light source. This feature can be used to clearly show the damage conditions of each layer of the laminate under different light source brightness; the hyperbolic tangent function fitting algorithm can retain the pixels with the same distribution pattern in the complete images of each damaged area of ​​the same layer, and remove the pixels with different distribution patterns in the complete images of different damaged areas, thereby reducing the error of the edge part of the complete image of a single damaged area.

[0041] Step 11: The high-definition images of the damaged area of ​​each layer of the laminate obtained in step 10 are superimposed together through the 3D drawing software Material Studio. The high-definition images of the damaged area of ​​each layer are represented by different colors, and finally a 3D damage model of the laminate is obtained.

[0042] As a preferred embodiment, the light source color is selected to be red, which has the strongest penetrability.

[0043] More preferably, the maximum brightness of the light source is 3000 nit, and the luminous flux is set to 1200 LM.

Claims

1. A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates, characterized in that: The following steps are involved: Step 1: Wipe the surface of the laminate, and then dry the moisture on the surface of the laminate; then fix the laminate to the laminate fixing plate, with the side of the laminate with the smaller damaged area facing downward; Step 2: Fix the camera on the camera stand and place it above the laminate; turn on the light source at the bottom of the laminate fixing plate and cover the light source with a light shield, so that the light from the light source passes through the light-transmitting port at the top of the light shield and the light-transmitting slot of the laminate fixing plate and irradiates the bottom of the laminate; observe the imaging effect of the camera through a computer, adjust the position of the camera stand on the camera fixing plate, align the shooting center of the camera with the center of the damaged area, and then fix the camera stand; Step 3: gradually increase the brightness of the light source, and observe the imaging effect of the camera through the computer until there is a light intensity difference between the boundary of the damaged area of ​​the laminate and other parts in the imaging photo transmitted by the camera to the computer, and save the light source brightness at this time as the initial light source brightness in the computer, and also save the imaging photo at this time in the computer; Step 4: Increase the brightness of the light source to a preset value, let it stand for a preset time, and save the image in the computer; Step 5: Repeat step 4 multiple times to obtain multiple imaging photos under different light source brightness; then, the computer calculates the brightness of all pixels in each imaging photo; Step 6: The computer uses an adaptive threshold binarization algorithm to process each image photo, so that the damaged area of ​​the layer with the largest damaged area of ​​the laminate in each image photo is processed into a bright part, and the rest of the laminate is processed into a dark part; Step 7: extracting damaged area images of each imaging photo under different light source brightness; Step 8: The damaged area images obtained in step 7 are fused to obtain a superimposed damaged area image; then, the edge of the superimposed damaged area image is fitted using a genetic algorithm; finally, the edge of the superimposed damaged area image after fitting is superimposed with the bright part of the superimposed damaged area image to obtain the final laminate overall damaged area image; Step nine: Use the OpenCV database to convert the imaging photos obtained in step five under different light source brightness into digital data, and then set a brightness separation range for the blocked parts of the damaged areas of each layer of the laminate from bottom to top, and separate the pixel data of the damaged area of ​​each layer of each imaging photo of the laminate that is larger than the damaged area of ​​the lower layer. Except for the bottom layer, the data of the damaged area of ​​each layer is the superposition of the pixel data of the damaged area of ​​the layer that is larger than the damaged area of ​​the lower layer, the pixel data of the damaged area of ​​each layer below the layer that is larger than the damaged area of ​​the lower layer, and the pixel data of the damaged area of ​​the bottom layer. The damaged area of ​​the bottom layer is the pixel with the highest brightness in the imaging photo; the data of the damaged area of ​​each layer of the laminate is converted into an image again to obtain a complete image of the damaged area of ​​each layer of the laminate in each imaging photo; Step 10: For the complete images of the damaged areas of each layer obtained in step 9, the complete images of the damaged areas of each group located on the same layer of the laminate are fitted using a hyperbolic tangent function fitting algorithm, and finally a high-definition image of the damaged area of ​​each layer of the laminate is obtained; Step 11: The high-definition images of the damaged area of ​​each layer of the laminate obtained in step 10 are superimposed together, and the high-definition images of the damaged area of ​​each layer are represented by different colors, and finally a 3D damage model of the laminate is obtained.

2. A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates according to claim 1, characterized in that: The color of the light source is red.

3. A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates according to claim 1, characterized in that: The maximum brightness of the light source is 3000nit, and the luminous flux is set to 1200LM.

4. A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates according to claim 1, 2 or 3, characterized in that: In step seven, the damaged area of ​​the imaging photo is extracted. The specific process is: use the image processing software Photoshop, click the lock polygon tool, use the mouse to select a lock so that the lock includes the bright part and the boundary between light and dark in the image, then use the "Select Inverse" command to select the undamaged area, click Delete, and save the separated damaged area image.

5. A nondestructive testing method based on the light transmittance characteristics of thermoplastic laminates according to claim 1, 2 or 3, characterized in that: The process of using genetic algorithm to fit the edge of the superimposed damaged area image in step 8 is as follows: ① Divide the edge of the superimposed damaged area image into edge regions to obtain an initial population; ② Calculate the fitness value f(T) = w1(T)*w2(T)*[u1(T)-u2(T)] for each edge region. 2 , where T represents the gray value threshold, w1(T) represents the number of pixels in the edge area whose gray values ​​are less than the gray value threshold, w2(T) represents the number of pixels in the edge area whose gray values ​​are greater than or equal to the gray value threshold, u1(T) represents the average gray value of all pixels in the edge area whose gray values ​​are less than the gray value threshold, and u2(T) represents the average gray value of all pixels in the edge area whose gray values ​​are greater than or equal to the gray value threshold; ③ Determine whether the fitness value of each edge area reaches the average fitness value The sum of the fitness values ​​of all edge regions divided by the total number of edge regions, select the edge regions whose fitness values ​​reach the average fitness value as the next generation population; ④ Calculate the average fitness value of the new generation population Select the edge area whose fitness value in the new generation population reaches the average fitness value as the next generation population; ⑤ Repeat step ④ for more than three times to obtain the edge of the superimposed damaged area image.

Images