A method for bridge crack detection and health status assessment based on computer vision

Abstract

The present invention discloses a method for detecting cracks in bridges and assessing health conditions based on computer vision, which is designed in the field of civil engineering technology and includes the following steps: S1, using a drone to take pictures of bridges, obtaining a picture set of bridges, taking a picture in the picture set as picture data input, and performing Gaussian denoising; S2, using CNN to determine whether there are cracks in the picture after Gaussian denoising, and graying the picture with cracks with a fixed threshold; S3, performing graphics calculation on the grayed picture to extract crack features; S4, assessing the health status of the bridge using crack features and material parameters, and obtaining a detection result after the assessment. The present invention utilizes computer vision technology, is easy to operate, saves costs, has higher detection accuracy and reliability, and can quickly assess the health status of bridge cracks.

Description

Technical Field

[0001] The invention relates to the technical field of civil engineering, and in particular to a method for bridge crack detection and health status assessment based on computer vision. Background Art

[0002] During the service life of a building, its cement parts often crack, which may affect the life and safety of the building. Currently, common detection methods mostly use infrared detection, sound wave detection and other technologies, which often cannot be automated or are costly, and are not conducive to daily maintenance and monitoring. Traditional manual bridge detection methods are expensive, and the detection results are mostly based on the experience of technicians. The detection efficiency is easily affected by the fatigue of the inspector, so it is highly subjective and the accuracy is uneven. For bridges in more remote locations, manual detection requires the use of some auxiliary facilities, and for bridges across canyons, turbulence and other locations, manual detection methods have high safety risks.

[0003] At present, the automatic detection method based on computer vision is used. Its main steps are to collect pictures of building cement surfaces and number them, and then detect them through morphological methods such as edge detection and threshold segmentation. However, there are many buildings with cement walls and many large buildings also use cement walls, which poses a great challenge to the manual collection of pictures. In addition, this technical method cannot calculate the length, slope, and area of ​​the cracks, and can only identify whether there are cracks here. By combining drone technology, it is possible to realize automatic collection of pictures based on manual detection, but it still cannot solve the problem of being unable to calculate the specific length and slope of the cracks. Summary of the invention

[0004] In order to solve the problems of high cost, strong subjectivity, low accuracy and low safety of traditional bridge detection technology, the present invention proposes a bridge crack detection and health status assessment method based on computer vision to solve the above problems.

[0005] The present application discloses a method for bridge crack detection and health status assessment based on computer vision, comprising the following steps:

[0006] S1. Use a drone to take pictures of bridges to obtain a picture set of bridges, take a picture from the picture set as picture data input, and perform Gaussian denoising;

[0007] S2, using CNN to determine whether there are cracks in the image after Gaussian denoising, and graying the image with cracks with a fixed threshold;

[0008] S3, performing graphics calculation on the grayscaled image to extract crack features;

[0009] S4. Use crack characteristics and material parameters to evaluate the health status of the bridge and obtain the test results after evaluation.

[0010] Preferably, S2 comprises the following steps:

[0011] CNN is used to extract image information and make a binary classification to determine whether the image is a crack image. The CNN uses a 4-layer convolutional neural network and a 4-layer pooling layer structure to extract information, and uses a dropout layer to temporarily discard nodes from the network with a probability of 0.1 during training to prevent overfitting.

[0012] Preferably, the crack characteristics in S3 include crack area, crack length and crack width.

[0013] Preferably, the calculation method of the crack area is:

[0014] Only the cracked part is retained in the grayscale image, and the black pixels in the image are counted to calculate the area of ​​the crack;

[0015] The method for calculating the crack length is as follows: obtaining the coordinates of the first black pixel point appearing from left to right and from top to bottom in the grayscale image as the starting point of the crack, obtaining the coordinates of the last black pixel point appearing as the end point of the crack, and calculating the length of the crack using the Pythagorean theorem;

[0016] The crack width is calculated as follows: first, the crack inclination θ in the vertical direction is calculated according to the cosine theorem, then the black pixels in each row of the image are counted, the average horizontal width wl of each row is calculated, and the crack width is calculated with the help of the inclination θ.

[0017] Preferably, the assessment of the health status of the bridge includes the assessment of the bending state of the reinforced concrete simply supported beam based on normal cracks and the assessment of the shear state of the reinforced concrete simply supported beam based on oblique cracks.

[0018] Preferably, the flexural state assessment of reinforced concrete simply supported beams based on normal cracks comprises the following steps:

[0019] Calculate the cracking moment using the following formula:

[0020] M cr =σ s A s (h 0 -x c / 3)+f t b(hx c )(2 / 3x c +hx c / 2);

[0021] Among them, σs is the steel bar stress, A s is the cross-sectional area of ​​the steel bar, h 0 is the difference between the section height and the distance from the reinforcement force point to the edge, x c is the distance between the crack and the edge of the beam, f t is the tensile strength of concrete, b is the section width, and h is the section height;

[0022] The calculation formula for calculating the bending moment of the cracked beam is:

[0023]

[0024] Among them, w 1 is the width of the first crack, w 2 is the width of the second crack, l is the distance between the first crack and the second crack, and f c is the compressive strength of concrete, E s is the elastic modulus of the steel bar, ε i1 It is the quotient of twice the tensile strength of concrete and the elastic modulus of concrete;

[0025] Calculate the ultimate bending moment using the following formula:

[0026]

[0027] Among them, f y It represents the yield strength of steel bar;

[0028] The bending state of the reinforced concrete simply supported beam with cracks is evaluated by comparing the cracking bending moment, the beam bending moment after cracking and the ultimate bending moment.

[0029] Preferably, the shear state assessment of a reinforced concrete simply supported beam based on an oblique crack comprises the following steps:

[0030] Calculate the cross-sectional shear force of the oblique crack using the following formula:

[0031] V=f 2 bj d cosθ·sinθ+f 1 bj d cosθ·sinθ;

[0032] Among them, f 1 is the principal tensile stress in cracked concrete, f 2 is the principal compressive stress of cracked concrete, j d is the length of the internal force arm, θ is the inclination of the crack in the vertical direction;

[0033] Calculate the shear bearing capacity using the following formula:

[0034]

[0035] Where v is the cross-sectional shear stress, A sv is the cross-sectional area of ​​the stirrup, f v is the design value of shear strength, s is the spacing between grip bars;

[0036] The shear state of reinforced concrete simply supported beams with oblique cracks is evaluated by comparing the cross-sectional shear force and shear bearing capacity of the oblique cracks.

[0037] Beneficial effects of the present invention:

[0038] (1) Easy to operate. The present invention uses computer vision technology to realize automatic recognition and parameter extraction of bridge cracks, which greatly simplifies the detection operation process, reduces the need for manual intervention, reduces the difficulty of operation, and improves safety;

[0039] (2) Cost saving. Traditional bridge health testing requires a lot of manpower and material resources, but the present invention saves human resources and reduces testing costs through automated processing and batch processing functions;

[0040] (3) High precision. The present invention uses advanced algorithms to process crack images, which can accurately extract crack width, length, area, inclination angle and other parameters, and has higher detection accuracy and reliability;

[0041] (4) Rapid assessment. With the bridge crack health assessment method of the present invention, users can quickly obtain the mechanical parameters of the crack image and the health assessment report, which provides a timely reference for the maintenance and repair of the bridge, reduces the risk of accidents, and prolongs the service life of the bridge. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of a method for bridge crack detection and health status assessment based on computer vision according to an embodiment of the present invention;

[0043] Figure 2 A schematic diagram of a CNN network structure according to an embodiment of the present invention;

[0044] Figure 3 A schematic diagram of graying a dark background image according to an embodiment of the present invention;

[0045] Figure 4 A schematic diagram of graying a light-colored background image according to an embodiment of the present invention;

[0046] Figure 5 A schematic diagram of performing graphics calculations according to an embodiment of the present invention;

[0047] Figure 6 A stress-strain distribution diagram of a critical state cross section of an embodiment of the present invention;

[0048] Figure 7 A schematic diagram of a normal cross-section crack generated in the mid-span of a reinforced concrete beam according to an embodiment of the present invention;

[0049] Figure 8 A simplified diagram of single reinforcement cross section calculation according to an embodiment of the present invention;

[0050] Fig. 9 This is a stress-strain distribution diagram of the destruction stage of an embodiment of the present invention;

[0051] Fig.10 is a unit average stress-strain diagram of an embodiment of the present invention;

[0052] Fig.11 Schematic diagram of cross-sectional balance relationship of an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and examples.

[0054] The present application embodiment discloses a bridge crack detection and health status assessment method based on computer vision, such as Figure 1 As shown, the following steps are included:

[0055] S1. Use a drone to take pictures of bridges to obtain a picture set of bridges, take a picture from the picture set as picture data input, and use computer technology to perform Gaussian denoising.

[0056] S2. Use the convolutional neural network (CNN) in computer vision technology to determine whether there are cracks in the image after Gaussian denoising. If there are no cracks, take another image and repeat the above operation; if there are cracks, grayscale the image with a fixed threshold.

[0057] The network structure of CNN is as follows Figure 2 As shown in the figure, CNN is used to extract image information and make a binary classification to determine whether the image is a crack image. CNN uses a 4-layer convolutional neural network and a 4-layer pooling layer (MaxPool) structure to extract information, and uses a dropout layer to temporarily discard nodes from the network with a probability of 0.1 during training to prevent overfitting.

[0058] In common crack images, the background color of the image, i.e. the color of the concrete itself, is not unique, but the color of the crack is relatively stable. Since cracks are fault structures with poor light reflection, in most images the cracks appear darker in color and are the darkest part of the entire image. Figure 3 As shown, the light background is grayed out as Figure 4 shown.

[0059] S3. Use OpenCV in computer vision technology to perform graphics calculations on the grayscale images to extract crack features, which include crack area, crack length and crack width.

[0060] The calculation method of the crack area is as follows: only the cracked part is retained in the grayscale image, the black pixels in the image are counted, and the crack area is calculated;

[0061] The calculation method of the crack length is as follows: analyze the image, obtain the coordinates of the first black pixel point appearing from left to right and from top to bottom in the grayscale image as the starting point of the crack, obtain the coordinates of the last black pixel point appearing as the end point of the crack, and use the Pythagorean theorem to calculate the length of the crack;

[0062] The calculation method of the crack width is as follows: first, the inclination θ of the crack in the vertical direction is calculated according to the cosine theorem, and then the black pixels in each row of the image are counted to calculate the average horizontal width wl of each row, and the width of the crack is calculated with the help of the inclination θ.

[0063] S4. Use crack characteristics and material parameters to evaluate the health status of the bridge and obtain the test results after evaluation.

[0064] According to the inclination θ, cracks can be divided into normal cracks and oblique cracks. For normal cracks, the bearing capacity of the structure affected by double cracks is calculated. The widths of the two cracks can be calculated through steps S1-S3, and the bearing capacity of the cracks on the overall structure can be determined by combining with other manually input concrete properties. For oblique cracks, the bearing capacity of the structure affected by single cracks is calculated. The inclination of the crack calculated above can be automatically obtained, and the bearing capacity of the cracks on the overall structure can be determined by combining with other manually input concrete properties.

[0065] Concrete beams often develop defects such as cracks under increasing loads. Defect features such as crack width and angle extracted by computer can be used to establish a mechanical relationship model to evaluate the health status of the structure and the relationship between internal force and ultimate bearing capacity. By establishing a bending state assessment model based on positive crack characteristics and a shear state assessment model based on oblique crack characteristics, the state of the component can be assessed, including the bending state assessment of reinforced concrete simply supported beams based on positive cracks and the shear state assessment of reinforced concrete simply supported beams based on oblique cracks. The reinforced concrete beam can be divided into elastic and inelastic stages from the beginning of stress to the final destruction, in which the concrete will experience elastic deformation, plastic deformation, cracks, crack expansion, and destruction.

[0066] like Figure 6The figure shows the stress-strain distribution diagram of the critical state section. When the load increases, the concrete begins to produce plastic deformation, and the growth rate of deformation is greater than the growth rate of stress, especially in the tensile area. Therefore, the stress image in the tensile area is a curve near the neutral axis, and most of the rest is a straight line. Finally, it is simplified as a rectangle; while in the compressive area, due to the small compressive stress, its compressive stress image is approximately a triangle. At this time, the concrete beam reaches the critical state of cracking, that is, the stress and strain at the lower edge of the tensile zone reaches the tensile strength limit of the concrete. At this time, the internal force of the section is the cracking moment M cr In this stage, the tensile zone of the concrete is close to cracking, and its stress reaches the standard value of tensile strength f t , becomes plastic state, and exerts greater plasticity, so the elastic-plastic modulus E can be approximated ct ≈0.5E c , E c is the initial elastic modulus of concrete; the strain of the steel bar is basically the same as the strain of the tensile edge of the concrete, that is, ε s =ε c . From the equilibrium condition, we can get:

[0067] σ s A s +f t b(hx c )=0.5bx c σ c ;

[0068] M cr =σ s A s (h 0 -x c / 3)+f t b(hx c )(2 / 3x c +hx c / 2);

[0069] Among them, σ s is the steel bar stress, a s is the cross-sectional area of ​​the steel bar, h 0 is the difference between the section height and the distance from the reinforcement force point to the edge, x c is the distance between the crack and the edge of the beam, f t is the tensile strength of concrete, b is the section width, h is the section height, σ c is the concrete stress.

[0070] The flexural state assessment of reinforced concrete simply supported beams based on normal cracks includes the following steps:

[0071] First, the relationship between the width of the cracks (positive cracks) in the positive section of the cracked concrete beam, the crack spacing and the stress of the steel bars is explored. Figure 7The figure shows the cross-sectional cracks generated in the mid-span of a reinforced concrete beam. Assume that the width of the first crack is w 1 , the width of the second crack is w 2 , the distance between the two cracks is l, and the strain of the concrete at the crack is equal to the strain of the steel bar at the critical point of cracking ε s The geometric relationship and constitutive relationship are:

[0072]

[0073] Among them, E s is the elastic modulus of the steel bar, ε i1 It is the quotient of twice the tensile strength of concrete and the elastic modulus of concrete;

[0074] Bending moment level M and ultimate bending bearing capacity M of reinforced concrete beam members in the service stage u Existence M / M u =0.5~0.7. To facilitate the analysis of the relationship between the beam bending moment and the tensile reinforcement stress, it is assumed that the cross section is a plane. The reinforcement stress should conform to f y is the yield strength of the steel bar; at the concrete crack section, the tensile strength of the concrete is 0; assuming that the concrete compressive stress is less than the axial compressive strength f c , compressive strain ε c ≤ε 0 =0.002,ε 0 is the ultimate compressive strain. The compressive stress-strain relationship of concrete is taken as σ u =f c [1-(1-ε c / ε 0 ) 2 ].

[0075] like Figure 8 The figure shows the calculation diagram of a single reinforcement section, and its equilibrium relationship is:

[0076] D=σ s A s ;

[0077] M=σ s A s (h 0 -c);

[0078] in, is the resultant compressive stress in the concrete compression zone, It is the distance from the point of action of the resultant compressive stress D in the compression zone of the concrete to the upper edge of the cross section.

[0079] Let λ = ε c / ε 0 , based on the double cracks, the bending moment of the cracked beam is calculated using the following formula:

[0080]

[0081] After analysis, it is found that the value of λ has little effect on the beam bending moment, and λ is greater than 0 and less than or equal to 1. Here, λ=0.5 is taken, and the formula is simplified to:

[0082]

[0083] When the load continues to increase, the stress of the steel bar increases rapidly. When the failure stage is approaching, the concrete in the compression zone is in the elastic-plastic stage. When the failure occurs, the steel bar in the tension zone reaches the yield strength, and the concrete in the compression zone reaches the ultimate compressive strain. Here, the component is considered as an adaptive beam component.

[0084] The stress-strain distribution diagram at the failure stage is shown in Fig. 9 As shown, the equilibrium relationship is:

[0085]

[0086] Among them, y 0 It is the distance from the longitudinal center of the steel bar to the compressive edge.

[0087] From this, the ultimate bending moment can be obtained as:

[0088]

[0089] y is the integral variable, that is, from 0 to y 0 The integral range of .

[0090] Since the concrete in the compression zone has reached the ultimate compressive strain, the strain at the outer edge of the compression concrete has exceeded. 0 =0.0033,ε c =0.002. Among them:

[0091] After finishing, we can get:

[0092]

[0093] Among them, f y Represents the yield strength of steel bars.

[0094] Therefore, the bending state of the reinforced concrete simply supported beam with a positive crack can be evaluated by comparing the cracking bending moment, the post-crack beam bending moment and the ultimate bending moment. That is, if the post-crack beam bending moment M is not greater than the cracking bending moment M cr , and M is not greater than the ultimate bending moment M u , the evaluation result is that the bridge is not damaged and the bridge is in good health; otherwise, the evaluation result is that the bridge is damaged and the bridge is in poor health.

[0095] The shear state assessment of reinforced concrete simply supported beams based on diagonal cracks includes the following steps:

[0096] Construct the relationship between the inclination θ of the oblique crack and the bearing capacity V u The mechanical relationship is obtained by calculating the mid-span bending moment to obtain the corresponding shear force V of the oblique crack section. The shear state of the simply supported beam is evaluated by comparing the bearing capacity limit with the shear force. At the same time, in order to simplify the calculation, combined with the modified pressure field theory, it is assumed that the angles of the principal compressive stress and principal compressive strain of the concrete are both inclinations. In this embodiment, only the low-reinforced section is considered. When the component reaches the ultimate failure, the concrete is not crushed but the stirrups are pulled to the yield state; the bond slip between the concrete and the steel bars is not considered.

[0097] It is assumed that the oblique cracks are evenly distributed and parallel to each other, and that the concrete between the cracks conforms to its own constitutive relationship; it is assumed that the steel bars in the concrete are evenly distributed, and the shear stress of the steel bars is not considered; the applied shear stress and normal stress are also evenly distributed. The shear bearing capacity of the cross section is calculated by the inclination θ of the oblique cracks obtained through actual detection, and the calculation method is as follows: Since most of the longitudinal bars and stirrups in reinforced concrete are steel bars with obvious yield points, in order to simplify the calculation, in this embodiment, the constitutive relationship of the stress and strain of the steel bars is simplified to a linear model, such as Fig.10 As shown, f cy is the stress in the y direction of concrete, f cx is the stress in the x direction of concrete, and its expression is:

[0098] When ε s ≤ε y When s =E s ε s ;

[0099] When ε y ≤ε s ≤ε sh When s =f y ;

[0100] Among them, ε s is the steel bar strain, ε y is the yield strain in the y direction, ε sh is the ultimate tensile strain of the steel bar, and in this embodiment, the value is 0.01.

[0101] When the inclined section stirrups reach yield, it is considered that the inclined section has reached the failure state. According to the stress Mohr circle, the relationship between the principal compressive and tensile stresses and the shear stress v of the section is:

[0102] (f 2 +f 1 ) = v·(tanθ+cotθ);

[0103] f 1 is the principal tensile stress in cracked concrete, f2 is the principal compressive stress in cracked concrete.

[0104] Since it is assumed that the steel bar does not produce bond slip, the strains of the two are the same, that is:

[0105] ε sx =ε cx =ε x ,ε sy =ε cy =ε y ;

[0106] Among them, ε xx is the strain of the reinforcement in the x direction, ε sy is the strain of the steel bar in the y direction, ε cx is the strain of concrete in the x direction, ε cy is the strain of concrete in the y direction, ε x is the yield strain in the x direction.

[0107] From the Wagner compatibility equation we get:

[0108]

[0109] And ε 1 +ε 2 =ε x +ε y , ε 1 is the tensile strain, ε 2 is the compressive strain;

[0110] Each strain of shear cracked concrete is the average strain of several cracks. x and ε y Pull is positive, ε 1 Pull is positive, ε 2 The pressure is negative, so

[0111] The constitutive relationship expression of the average compressive stress-compressive strain of concrete in the modified pressure field theory is:

[0112]

[0113] Among them, f 2max is the ultimate compressive stress of concrete, and Here ε 1 It is a negative value, with a value of -0.002. This constitutive relationship takes the tensile and compressive softening effect of concrete into consideration.

[0114] The principal tensile stress f of cracked concrete in the modified pressure field theory 1 With tensile strain ε 1 The relationship is divided into two situations. Before the concrete cracks, it is in the elastic stage:

[0115] f 1 =E c ε 1 ;

[0116] After concrete cracking:

[0117]

[0118] Among them, take E c =2f c / ε 0 , ε 0 =0.002; α 1 is the bonding factor between steel bars and concrete. In view of the fact that deformed steel bars are commonly used at present, the value is taken as 1.0; α 2 The load type factor is 0.7 in practical applications.

[0119] like Fig.11 The figure shows a schematic diagram of the cross-sectional equilibrium relationship. From the equilibrium condition in the cross-sectional y direction, the cross-sectional shear force of the oblique crack can be obtained as follows:

[0120] V=f 2 bj d cosθ·sinθ+f 1 bj d cosθ·sinθ;

[0121]

[0122] From the internal force balance of stirrups, we can get:

[0123] A sv f v =(f 2 sin 2 θ-f 1 cos 2 θ)bs;

[0124] Among them, A sv is the cross-sectional area of ​​the stirrup, f v is the design value of shear strength, and s is the spacing between the reinforcing bars.

[0125] When both sides of the equation are equal, the calculation can continue. Otherwise, increase the ε 1 value, and recalculate;

[0126] According to the above formula, the shear bearing capacity is derived and the calculation formula is:

[0127]

[0128] When shear failure is reached, the stirrups yield, that is, f v=f y .

[0129] Therefore, the shear bearing capacity can be calculated by the characteristics of the inclined crack of the reinforced concrete simply supported beam. The cross-sectional shear force is calculated by using the mid-span bending moment of the beam, and the shear bearing capacity is compared with the cross-sectional shear force to evaluate the shear state of the reinforced concrete simply supported beam. That is, if the cross-sectional shear force V is not greater than the shear bearing capacity V u , the evaluation result is that the bridge is not damaged and the bridge is in good health; if the cross-sectional shear force V is greater than the shear bearing capacity V u , the evaluation result is that the bridge is damaged and the bridge health status is poor.

[0130] The health status assessment of bridge cracks is completed by comprehensively evaluating the bending state of reinforced concrete simply supported beams based on normal cracks and the shear state assessment of reinforced concrete simply supported beams based on oblique cracks.

[0131] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A bridge crack detection and health status assessment method based on computer vision, characterized in that: The following steps are involved: S1. Use a drone to take pictures of bridges to obtain a picture set of bridges, take a picture from the picture set as picture data input, and perform Gaussian denoising; S2, using CNN to determine whether there are cracks in the image after Gaussian denoising, and graying the image with cracks with a fixed threshold; S3, performing graphics calculation on the grayscaled image to extract crack features; S4. Evaluate the health status of the bridge using crack characteristics and material parameters, and obtain a test result after evaluation, wherein the evaluation of the health status of the bridge includes evaluating the bending state of the reinforced concrete simply supported beam based on positive cracks and evaluating the shear state of the reinforced concrete simply supported beam based on oblique cracks; The flexural state assessment of reinforced concrete simply supported beams based on positive cracks comprises the following steps: Calculate the cracking moment using the following formula: M cr =σ s A s (h0-x c / 3)+f t b(h-x c )(2 / 3x c +h-x c / 2); Among them, σ s is the steel bar stress, A s is the cross-sectional area of ​​the steel bar, h0 is the difference between the cross-sectional height and the distance from the steel bar resultant force point to the edge, x c is the distance between the crack and the edge of the beam, f t is the tensile strength of concrete, b is the section width, and h is the section height; The calculation formula for calculating the bending moment of the cracked beam is: Where w1 is the width of the first crack, w2 is the width of the second crack, l is the distance between the first crack and the second crack, and f c is the compressive strength of concrete, E s is the elastic modulus of the steel bar, ε i1 It is the quotient of twice the tensile strength of concrete and the elastic modulus of concrete; Calculate the ultimate bending moment using the following formula: Among them, f y It represents the yield strength of steel bar; The bending state of reinforced concrete simply supported beams with cracks can be evaluated by comparing the cracking bending moment, the beam bending moment after cracking and the ultimate bending moment. The shear state assessment of reinforced concrete simply supported beams based on oblique cracks comprises the following steps: Calculate the cross-sectional shear force of the oblique crack using the following formula: V=f2bj d cosθ·sinθ+f1bj d cosθ·sinθ; Where f1 is the principal tensile stress of cracked concrete, f2 is the principal compressive stress of cracked concrete, and j d is the length of the internal force arm, θ is the inclination of the crack in the vertical direction; Calculate the shear bearing capacity using the following formula: Where v is the cross-sectional shear stress, A sv is the cross-sectional area of ​​the stirrup, f v is the design value of shear strength, s is the spacing between grip bars; The shear state of reinforced concrete simply supported beams with oblique cracks is evaluated by comparing the cross-sectional shear force and shear bearing capacity of the oblique cracks.

2. The method for bridge crack detection and health status assessment based on computer vision according to claim 1, characterized in that: The S2 comprises the following steps: CNN is used to extract image information and make a binary classification to determine whether the image is a crack image. The CNN uses a 4-layer convolutional neural network and a 4-layer pooling layer structure to extract information, and uses a dropout layer to temporarily discard nodes from the network with a probability of 0.1 during training to prevent overfitting.

3. The method for bridge crack detection and health status assessment based on computer vision according to claim 2 is characterized in that: The fracture characteristics in S3 include fracture area, fracture length and fracture width.

4. The method for bridge crack detection and health status assessment based on computer vision according to claim 3 is characterized in that: The calculation method of the crack area is: Only the cracked part is retained in the grayscale image, and the black pixels in the image are counted to calculate the area of ​​the crack; The method for calculating the crack length is as follows: obtaining the coordinates of the first black pixel point appearing from left to right and from top to bottom in the grayscale image as the starting point of the crack, obtaining the coordinates of the last black pixel point appearing as the end point of the crack, and calculating the length of the crack using the Pythagorean theorem; The crack width is calculated as follows: first, the crack inclination θ in the vertical direction is calculated according to the cosine theorem, then the black pixels in each row of the image are counted, the average horizontal width wl of each row is calculated, and the crack width is calculated with the help of the inclination θ.

Images