Hollow slab bridge regional distribution rigidity identification method based on visual measurement

Abstract

The present application belongs to the technical field of building structure health monitoring, and particularly relates to a hollow slab bridge regional distribution stiffness identification method based on visual measurement, which can realize regional stiffness identification of a hollow slab bridge. The specific steps include: under the action of random traffic load, vehicle reference point positioning and vehicle information identification are performed by using a bridge vehicle flow monitoring camera, and in combination with toll station vehicle weight data, the position and size of the bridge surface vehicle load at any time are obtained; while identifying the bridge surface load, the angles of the nodes of each test unit in the test bridge region are synchronously monitored by using an online camera under the bridge; according to the position and size of the bridge surface vehicle load at any time, the matrix displacement method based on the beam grillage model is used to calculate the bending moment envelope area of each stiffness identification region at any time; according to the monitored unit node angles, the curvature envelope area of each stiffness identification region is calculated; according to the quotient of the bending moment envelope area and the curvature envelope area of each region, and taking the average in the time scale, the uniform stiffness of each test unit is obtained; according to the uniform stiffness of each test unit, the standardized stiffness damage index is used to identify the structural damage; if the identified structure is undamaged, the measurement is ended; if the identified structure is damaged, the internal force redistribution effect after the regional stiffness damage is considered for the damaged structure, the beam grillage model parameters are updated by using the iteration method until convergence, and the regional stiffness under the damage is obtained. The present method considers that the hollow slab bridge has a large width-span ratio and obvious lateral distribution effect, the matrix displacement method based on the beam grillage model is used to calculate the bending moment distribution of the structure, which is fast, convenient, high in test precision, and suitable for the rapid test of the regional distribution stiffness of the hollow slab bridge.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of building structure health monitoring, and particularly relates to a hollow slab bridge regional distribution stiffness identification method based on visual measurement, which can realize regional stiffness identification of a hollow slab bridge. BACKGROUND

[0002] As an important part of modern traffic network, bridges may prematurely appear fatigue damage and even structural damage under long-term unreasonable vehicle action. Using structural monitoring technology to find structural diseases and hidden dangers in time is expected to solve the structural service safety problem and provide strong support for realizing long service life. In the design and construction process of small and medium span bridges in China, prefabricated reinforced concrete beam bridges are widely used due to their simple structure and good mechanical properties. Hollow slabs have unique advantages such as small self-weight, simple prefabrication and convenient construction, so in the construction of small and medium span bridges, prefabricated hollow slab beam bridges become the first choice and account for more than 50% of small and medium span beam bridges in China. Influenced by design, construction, natural erosion and operating environment, etc., the prestressed concrete hollow slab beam bridge built will gradually appear diseases such as material degradation, cracks and structural damage as the service time is prolonged.

[0003] In the field of structural identification, if the input and output information of the structure can be effectively used, it can provide strong support for the identification, evaluation and maintenance of bridge structures. The non-contact measurement advantage based on visual technology has the potential to realize the acquisition of structural load input information while acquiring real-time response output measurement based on online cameras when monitoring the space-time distribution of bridge deck vehicle weight.

[0004] Accurate identification of bridge essential characteristic information can provide important structural health information, which helps to find potential material degradation, structural crack disease problems in time and take corresponding repair and reinforcement measures to ensure the safe operation of the bridge. Existing health monitoring research has made certain progress in the aspects of multi-vision information perception such as bridge vehicle flow space-time distribution acquisition (input) based on visual technology and structural response monitoring (output) based on optical measurement, but there are still limitations in the aspects of using health monitoring data to acquire structural essential characteristic information and evaluating the service condition of hollow slab beam bridges.

[0005] The environment vibration test only depending on the "output" information can only identify the basic modal parameters, which have weak correlation with the structural performance and cannot effectively support the performance evaluation of the structure; the impact vibration test with the known "input-output" information can identify the deep-level parameters such as the structural flexibility matrix, but the need for the measurement of the characteristics of the input impact load limits the use thereof to be closed to traffic, which brings inconvenience to the monitoring of the large-scale and wide small and medium-sized bridges; the existing mobile load test based on the car test utilizes the quasi-static influence line index to reflect the overall stress performance of the structure, but also has the limitation of closed traffic. The vehicle load data on the bridge are used as the structural input, and the structural deformation data are used as the structural response output, so that the traffic is not closed, and the deep-level parameters of the structure can be obtained, and therefore, the development of the hollow slab bridge damage identification method based on the vehicle flow identification and the structural displacement has the practical engineering significance and the clear engineering demand. SUMMARY

[0006] The technical problem solved by the present application is that the present application discloses a hollow slab bridge regional distribution stiffness identification method based on visual measurement, which has the advantages of fast and convenient measurement and non-contact, overcomes the problem that the existing method needs to close the traffic, and has a good prospect of wide application in the hollow slab bridge regional stiffness identification.

[0007] The present application adopts the following technical scheme:

[0008] A hollow slab bridge regional distribution stiffness identification method based on visual measurement comprises the following steps:

[0009] S1, under the action of random traffic load, the vehicle reference point positioning and vehicle information identification are performed by using the vehicle flow monitoring camera on the bridge, and the position and size of the vehicle load on the bridge at any time are obtained in combination with the vehicle weight data of the toll station;

[0010] S2, while the bridge load is identified, the rotation angles of the nodes of each test unit in the test bridge region are synchronously monitored by using the online camera.

[0011] S3, according to the size and position of the traffic load, the matrix displacement method based on the grillage model is used to calculate the bending moment envelope area of each stiffness identification region at any time.

[0012] S4, according to the rotation angles of the unit nodes, the curvature envelope area of each stiffness identification region is calculated.

[0013] S5, the quotient of the bending moment envelope area and the curvature envelope area of each region is calculated, and the average value is taken in the time scale to obtain the uniform stiffness of each test unit.

[0014] S6, according to the uniform stiffness of each test unit, the standardized stiffness damage index is calculated to identify the structural damage.

[0015] S7, for the damaged structure, considering the internal force redistribution effect after the regional stiffness damage, the iterative method is used to update the parameters of the grillage model until convergence, and the regional stiffness under damage is obtained.

[0016] Further, in step S1, the vehicle reference point positioning and vehicle information recognition are performed by the bridge vehicle flow monitoring camera, and the position and size of the bridge vehicle load at any time are obtained by combining the toll station vehicle weight data, including the following sub-steps:

[0017] S11, Yolov5s is used to detect vehicles on the whole bridge picture, and for each frame of picture, the left lower point of the vehicle head (specifically: the leftmost point among the three lowest feature points on the mask contour) is recognized as the reference point by using the recognized vehicle mask;

[0018] S12, vehicle tracking is performed by using template matching, the correlation coefficient between each vehicle subgraph in the next frame and each vehicle subgraph in the previous frame is calculated to form a correlation matrix, and the peak value of each row is found from the correlation matrix, and the row and column representing the position of the peak value represent the same vehicle. Each vehicle is assigned a label when it first enters the field of view, and the vehicle matching with this vehicle in the next frame is assigned the same label. Through the cycle, the label of this vehicle in all frames is the same, realizing vehicle tracking and reference point matching;

[0019] S13, camera calibration is performed by using the pixel coordinate information of the feature points on the picture and the known actual spatial coordinate information to obtain the homography matrix from the image coordinate system to the bridge world coordinate system; and the reference point position of each vehicle in the bridge coordinate system at any time is calculated through the homography matrix;

[0020] S14, when the vehicle enters the vehicle information recognition area, Yolov5s is used to recognize the mask of the vehicle and the wheel. The lowest point of the mask contour is taken as the reference point of the vehicle and the wheel; and the position of the wheel and vehicle reference point in the bridge coordinate system is further calculated according to the homography matrix.

[0021] S15, at the same time, the edge frame of the license plate is recognized by using YOLOv5s; and the characters of the license plate are further recognized by using LPRNet, so as to correspond to the vehicle weight data.

[0022] S16, the vehicle reference point position obtained in S13 and the position of the wheel relative to the reference point obtained in S14 are combined to obtain the load position; and the vehicle weight data obtained in S15 is combined to obtain the load size.

[0023] Furthermore, in step S2, the following method is used to synchronously monitor the rotation angle of each test unit node in the test bridge area using an online camera: the array target image sequence with multiple feature point position constraints is used as visual input to construct the geometric relationship between the camera and the coordinate system of the moving target to calculate the bridge rotation angle, and the robustness of the rotation angle monitoring is improved by using a centroid array adaptive multi-wave gate tracking and coordinate system transformation matrix purification and optimization.

[0024] Furthermore, in step S3, based on the magnitude and location of the traffic load, the bending moment envelope area of ​​each stiffness identification region at any given time is calculated using the matrix displacement method based on the beam grid model, including the following steps:

[0025] S31 uses the beam grid method to simplify the hollow slab beam bridge. The transverse and longitudinal stiffness of the hollow slab are concentrated using transverse and longitudinal beam grids, with hinges at joints and supports. Each transverse connection consists of two members, each rigidly connected to its corresponding longitudinal member, and the two members are hinged to each other.

[0026] S32, for each member, establish a local coordinate system, denoted as e in the entire structure, connecting two nodes i and j. Now, with i as the origin and the direction from i to j as... The positive direction of the axis, vertically upward is Establish a right-handed Cartesian coordinate system. In each member, neglect the axial force; assume that the nodes only translate along the y-direction and do not rotate about it. That is, the force component at the member ends that needs to be considered includes the bending moment in the x-direction at end i. shear force in the y direction Bending moment in the z-direction and the bending moment in the x direction at end j shear force in the y direction Bending moment in the z-direction According to the displacement method, the stiffness equation of the element is: In the formula, the force vector at the rod end and rod end displacement vector respectively Element stiffness matrix for

[0027]

[0028] Among them (GI) t ) e Let EI be the torsional stiffness of member e. e For the bending stiffness of member e, l e Let e ​​be the length of member e. In this method, since the matrix displacement method is only used to calculate the force distribution and does not need to output the structural deformation, the material elastic modulus can take any value. The bending moment of inertia of the section is I = dh. 3 / 12, torsional inertia moment I t = 0.33dh 3 ; for the transverse bar, since the hinge joint does not transmit torque, the transverse bar torsional stiffness does not affect the structural response, for convenience, set the transverse bar I t = 0, since the plate beam will produce elastic stage deflection in the longitudinal direction, but in the transverse direction it can be considered as rigid, that is, the transverse bar bending stiffness in the model is much larger than the longitudinal bar, which can be considered as infinite, here the cross-section bending inertia moment is assigned I = lh 3 / 36, to adapt to the calculation of the matrix displacement method.

[0029] S33, after obtaining the element stiffness matrix, use coordinate transformation to obtain the element stiffness matrix in the global coordinate system, and block, to facilitate the next step of calculation. The global coordinate system and the local coordinate system only rotate around the y axis, so the coordinate transformation matrix T is

[0030]

[0031] where α is the angle from the x axis to the y axis. For longitudinal beam grids, α = 0°; for transverse beam grids, α = 270°.

[0032] S34, assemble the global stiffness matrix from the element stiffness matrix. Set the beam grid method model to have n nodes, then the total stiffness matrix K has n 2 sub-blocks, each sub-block is a 3x3 order matrix, so K is a 3n x 3n order matrix. The main sub-block K ii is obtained by superimposing the main sub-blocks of each related element of node i, that is, where the sub-block K im , when i, m are related nodes, that is, the corresponding sub-block of the element connecting them, that is, when i, m are non-related nodes, it is a zero sub-block.

[0033] S35, after combining into the global stiffness matrix K, according to the original stiffness equation of the structure, F = KΔ, where F is the external force of each node, and α is the displacement of each node.

[0034] F = (F1 F2…F n ) T = (M x1 F y1 M z1 M x2 F y2 M z2 …M xn F yn M zn ) T

[0035]

[0036] Here, F i is the column vector of external force of node i, M xi is the column vector of external moment of node i, F yi is the column vector of external force of node i, M zi is the column vector of external moment of node i, Δ i is the column vector of displacement of node i, θ xi is the column vector of angular displacement of node i, v yi is the column vector of linear displacement of node i.

[0037] Before introducing the support conditions, K is a known 3n x 3n order matrix. F and Δ are unknown column vectors with 3n elements. Therefore, 3n constraints are needed to solve all the parameters in the original stiffness equation. For a hollow slab bridge composed of p slab beams, q groups of transverse beam grids are set, the structure has (p-1)q inter-beam hinge points, 2p support nodes, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joint. For the hinge nodes at the 2p beam end supports, θ xi = 0, v yi = 0, M zi = 0; for the hinge nodes corresponding to the (p-1)q transverse connections, let the nodes on both sides of the hinge be node k and node (k+1), then M xk = 0, M x(k+1) = 0, F yk + F y(k+1) = 0, M zk = 0, M z(k+1) = 0, v yk = v y(k+1) ; for the nodes at the pq transverse and longitudinal beam connections, M xi = 0, F yi = F E , M zi = M E , where F E and M E are the equivalent node loads obtained by distributing the bridge vehicle load on the nodes in S1, and the equivalent node load distribution method is as follows:

[0038] ​Since the width dimension of the hollow slab beam is much smaller than the length dimension, the concentrated force generated by the vehicle is considered to act on the longitudinal beam grid, i.e. the eccentric load on a single hollow slab beam is not considered. For the longitudinal member, first, the additional chain link and rigid arm are added to prevent the linear displacement and angular displacement of all nodes, at this time each unit has a fixed end force (hereinafter referred to as a fixed end force), and the additional counterforce and counterforce couple are on the additional chain link and rigid arm. According to the node balance, the values of these additional counterforce and counterforce couple are equal to the algebraic sum of the fixed end forces converging at the node. Then, the additional chain link and rigid arm are removed, i.e. the above-mentioned additional counterforce and counterforce couple are added to the node as loads after being reversed, these loads are called equivalent node loads of the original non-node loads, and the actual displacement of the node is obtained under the equivalent node load. The longitudinal member under the vehicle load is only subjected to the y-direction concentrated force F, and the fixed end force F F The calculation is as follows The equivalent node load at any node is the sum of the fixed end forces of all the members connected thereto,

[0039] The above 3n constraint conditions are combined with the original stiffness equation of the structure to obtain the actual displacement Δ of each node;

[0040] The member end internal force F of member e is calculated by the following formula e , including the x-direction member end bending moment y-direction member end shear force

[0041] z-direction member end bending moment i.e. F e is the sum of the fixed end force and the member end force generated under the action of the equivalent node load:

[0042]

[0043] where F Ei is the equivalent node load of node i, k e is the unit stiffness matrix in the global coordinate system, and Δe is the unit displacement matrix in the global coordinate system. (k e Δ e ) i The member end force at i end generated under the action of the equivalent node load. The x-direction member end bending moment is matched with the bending moment diagram to obtain the bending moment distribution inside the member; i.e. the bending moment at any length of each hollow slab beam; S36, after obtaining the bending moment at any length of each hollow slab beam, further according to calculate the bending moment envelope area of the region [x i , x i+1 ] on the nth hollow slab at time t.

[0044] Further, the curvature envelope area of each stiffness identification region is calculated according to the corner of the unit node in step S4, using the formula A C (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ), wherein θ(n,t,x i ) is the corner of the nth hollow slab at time t at x i , and A C (n,t,x i ) is the curvature envelope area of the region [x i , x i+1 ] of the nth hollow slab at time t.

[0045] Further, the uniform stiffness of the ith test unit A(n,i) of the nth hollow slab is calculated using the formula in step S5, wherein T is the total effective measurement time.

[0046] Further, the normalized stiffness damage index is calculated using the formula in step S6 to identify the damage, wherein is the uniform stiffness of the region A(n,i), μ is the average of the stiffness of all regions, and σ is the standard deviation of the stiffness of all regions.

[0047] Further, the parameters of the grillage model are updated using an iterative method to obtain the stiffness of the damage region in step S7, and the specific steps are as follows:

[0048] S71, the bending stiffness value under the assumption that the structure is intact is obtained, and the damage is identified, the stiffness value of the intact region is averaged and assigned to all intact regions of the longitudinal grillage; the expression is as follows:

[0049]

[0050] wherein represents the average uniform stiffness of the intact region under the assumption that the structure is intact, is the stiffness of the intact region used in the first iteration.

[0051] The bending stiffness value of the damage region is equal to the stiffness value of this region under the assumption that the structure is intact,

[0052]

[0053] wherein represents the uniform stiffness of the damage region (n,i) under the assumption that the structure is intact, is the stiffness of the damage region (n,i) used in the first iteration.

[0054] For torsional stiffness, it is considered that it varies proportionally with the bending stiffness;

[0055] S72, using the stiffness value obtained by S71, updating the unit stiffness matrix k e , repeating steps S34-S36 to obtain the bending moment envelope area (A M (n,t,x i )) loop 1; using the curvature envelope area A C (n,t,x i ), for each region, the uniform stiffness of the region (n,i) obtained by the first iteration is calculated

[0056]

[0057] S73, the stiffness value of the intact region is averaged again and assigned to all intact regions,

[0058]

[0059] wherein, is the average value of the uniform stiffness of the intact region obtained by the (i-1)th iteration, is the stiffness value of the intact region used in the ith iteration.

[0060] The bending stiffness value of the damaged region is equal to the stiffness value of this region calculated last time,

[0061]

[0062] wherein, is the uniform stiffness of the damaged region (n,i) obtained by the (i-1)th iteration, is the uniform stiffness of the damaged region (n,i) used in the ith iteration.

[0063] The application provides a hollow slab bridge regional distribution stiffness identification method based on visual measurement. First, the computer vision method is applied to identify the vehicle position and information, the bridge surface vehicle weight space-time distribution is obtained by combining the toll station vehicle weight data, and then the matrix displacement method is used to realize the bending moment envelope area calculation of each region; At the same time, the optical measurement is applied to realize the synchronous measurement of the structure rotation angle response, and the curvature envelope area is calculated accordingly. The quotient of the bending moment envelope area and the curvature envelope area is averaged in the time scale to obtain the stiffness of each test region. Thereafter, the standardized stiffness damage index is calculated to identify the damage, and for the damaged structure, the internal force redistribution effect is considered, the model parameters are updated by using the iteration method, and the regional stiffness under the damage is accurately quantified.

[0064] Advantages:

[0065] Firstly, the hollow slab bridge regional distribution rigidity identification method based on visual measurement solves the problems of traditional bridge multi-point rigidity measurement, such as the need of a large number of contact sensors, the need of closing traffic, and high test cost.

[0066] Secondly, the method considers the large width-span ratio of the hollow slab bridge and obvious lateral distribution effect, and utilizes the matrix displacement method based on the grillage model to calculate the bending moment distribution of the structure, so that the method is fast, convenient, high in test precision, and suitable for the rapid test of the regional distribution rigidity of the hollow slab bridge.

[0067] Thirdly, the method considers the internal force redistribution effect after the regional rigidity damage of the hollow slab bridge, and utilizes the iteration method to update the parameters of the grillage model, so that the regional rigidity under damage can be accurately quantified, and the method is suitable for the damage state evaluation of the bridge. DETAILED DESCRIPTION

[0068] Figure 1 It is a principle schematic diagram of the hollow slab bridge regional distribution rigidity identification method based on visual measurement of the embodiment of the present application.

[0069] Figure 2 It is a vehicle tracking schematic diagram based on computer vision.

[0070] Figure 3 It is a vehicle information identification schematic diagram based on computer vision.

[0071] Figure 4 It is a principle schematic diagram of the grillage method.

[0072] Figure 5 It is a principle schematic diagram of the equivalent load distribution.

[0073] Figure 6 It is a laboratory layout diagram.

[0074] Figure 7 It is a rigidity calculation display diagram under the perfect working condition.

[0075] Figure 8 It is a rigidity calculation display diagram under the damage working condition. DETAILED DESCRIPTION

[0076] The following embodiments can enable a person skilled in the art to more comprehensively understand the present application, but do not limit the present application in any way.

[0077] Reference Figure 1 The embodiment discloses a hollow slab bridge regional distribution rigidity identification method based on visual measurement.

[0078] S1, under the action of random traffic load, vehicle reference point positioning and vehicle information identification are performed by using a vehicle flow monitoring camera on the bridge, and combined with the vehicle weight data of a toll station, the position and size of the vehicle load on the bridge at any time are obtained.

[0079] S2, while identifying the bridge load, the corner of each test unit node in the test bridge area is monitored synchronously by using an online camera.

[0080] S3, according to the size and position of the traffic load, the matrix displacement method based on the grillage model is used to calculate the bending moment envelope area of each stiffness identification area at any time.

[0081] S4, according to the corner of the unit node, the curvature envelope area of each stiffness identification area is calculated.

[0082] S5, the quotient of the bending moment envelope area and the curvature envelope area of each area is taken, and the average value is taken in the time scale to obtain the uniform stiffness of each test unit.

[0083] S6, according to the uniform stiffness of each test unit, the standardized stiffness damage index is calculated to identify the structural damage.

[0084] S7, for the damaged structure, considering the internal force redistribution effect after the regional stiffness damage, the iterative method is used to update the grillage model parameters until convergence, and the regional stiffness under damage is obtained.

[0085] Specifically includes the following steps:

[0086] Step 1: Obtain the position and size of the bridge vehicle load

[0087] The traffic monitoring camera is erected above the bridge deck, and the camera direction is adjusted to completely cover the bridge deck. The data acquisition system collects traffic images and saves the collected image sequence files.

[0088] As shown in Figure 2 , YOLOv5s is used to detect vehicles on the whole bridge picture. For each frame of picture, the left lower point of the vehicle head (specifically: the leftmost point among the three lowest feature points on the mask contour) is used as the reference point by using the identified vehicle mask. Vehicle tracking is performed by template matching. The correlation coefficient between each vehicle subgraph in the next frame and each vehicle subgraph in the previous frame is calculated to form a correlation matrix. The peak value of each row is found in the correlation matrix, and the row and column represented by the peak value represent the same vehicle. Each vehicle is assigned a label when it first enters the field of view, and the vehicle matching in the next frame is assigned the same label. Through the cycle, the label of this vehicle is the same in all frames, realizing vehicle tracking and reference point matching. The pixel coordinate information of the feature points on the picture and the known actual space coordinate information are used for camera calibration to obtain the homography matrix from the image coordinate system to the bridge deck world coordinate system. The reference point position of each vehicle in the bridge deck coordinate system at any time is calculated through the homography matrix. As shown in Figure 3As shown, when the vehicle enters the vehicle information recognition area, the vehicle and wheel mask recognition are performed on it by using Yolov5s, the lowest point of the mask contour is taken as the reference point of the vehicle and the wheel; According to the homography matrix, the positions of the wheel and the vehicle reference point in the bridge coordinate system are further calculated. At the same time, the frame of the license plate is recognized by using YOLOv5s; The license plate characters are further recognized by using LPRNet, so as to correspond to the vehicle weight data. Combined with the position of the vehicle reference point and the position of the wheel relative to the reference point, the load position is obtained; Combined with the vehicle weight data, the load size is obtained.

[0089] Step 2: Node corner synchronization monitoring

[0090] The online camera is erected under the bridge, the direction and focal length of the online camera are adjusted, so that all the measuring points are imaged clearly in the field of view, the data acquisition system collects the corner monitoring image and saves the collected image sequence file, and the image sequence collected by the traffic monitoring camera on the bridge is aligned according to the time stamp.

[0091] When processing the image, the array target image sequence with multiple feature point positions is taken as the visual input, the geometric relationship between the camera and the moving target coordinate system is constructed to calculate the bridge corner, and a kind of centroid array adaptive multi-wave door tracking and coordinate system conversion matrix purification optimization is used to improve the robustness of the corner monitoring.

[0092] Step 3: Identification of the bending moment envelope area of the region

[0093] As shown in Figure 4 The beam grillage method is used to simplify the hollow slab beam bridge. The transverse and longitudinal stiffness of the hollow slab is concentrated by the transverse and longitudinal beam grillage, and the hinge joint and support are hinged. Each transverse connection is composed of two rod members, and each rod member is respectively hinged to its corresponding longitudinal member. For each rod member, a local coordinate system is established, and its number in the whole structure is e, which connects two nodes i and j. Now take i as the origin, and the direction from i to j as the positive direction of the axis, and the vertical upward direction as the axis, to establish a right-hand space rectangular coordinate system. In each rod member, the axial force is ignored; it is considered that the node only moves along the y direction and does not rotate around the y direction. That is, the rod end force components that need to be considered are the bending moment of the i end in the x direction, the shear force of the j end in the y direction, and the bending moment of the j end in the z direction.

[0094] According to the displacement method, the stiffness equation of the unit is where the rod end force vector and the rod end displacement vector​​ respectively element stiffness matrix is:

[0095]

[0096] where (GI t ) e is the torsional stiffness of member e, (EI) e is the flexural stiffness of member e, and l e is the length of member e. In this method, since the matrix displacement method is only used to calculate the distribution of forces, without the need to output the structural deformation, the material modulus of elasticity can take any value. The flexural inertia of the cross section I = dh 3 / 12, and the torsional inertia of the cross section I t = 0.33dh 3 ; for the transverse members, since the hinge joint does not transmit torque, the torsional stiffness of the transverse member does not affect the structural response, for the sake of convenience, set the transverse member I t = 0, since the plate girder will produce elastic stage deflection in the longitudinal direction, but in the transverse direction it can be considered as rigid, that is, the flexural stiffness of the transverse member in the model is much larger than that of the longitudinal member, which can be considered as infinite, here the flexural inertia of the cross section is assigned I = lh 3 / 36 to adapt to the calculation of the matrix displacement method.

[0097] After obtaining the element stiffness matrix, use to perform coordinate transformation to obtain the element stiffness matrix in the global coordinate system, and perform block division to facilitate the next calculation. The global coordinate system and the local coordinate system only rotate around the y-axis, so the coordinate transformation matrix T is

[0098]

[0099] where α is the angle from the x-axis to the y-axis. For longitudinal beam grids, α = 0°; for transverse beam grids, α = 270°.

[0100] Assemble the global stiffness matrix from the element stiffness matrix. Suppose there are n nodes in the beam grid method model, then the total stiffness matrix K has n 2 sub-blocks, each of which is a 3x3 order matrix, so K is a 3n x 3n order matrix. Among them, the main sub-block K ii is obtained by superimposing the main sub-blocks of each related element of node i, that is, where the sub-block K im , when i, m are related nodes, that is, the corresponding sub-block of the element connecting them, that is, when i, m are non-related nodes, it is a zero sub-block.

[0101] After the assembly of the global stiffness matrix K, according to the original stiffness equation of the structure, F = KΔ, where F is the external force of each node and Δ is the displacement of each node.

[0102] F = (F1 F2...Fn) n ) T = (M x1 F y1 M z1 M x2 F y2 M z2 ...M xn F yn M zn ) T

[0103]

[0104] Here, F i represents the external force column vector of node i, M xi , F yi , M zi are the external force couple in the x direction, the external force in the y direction, and the external force couple in the z direction acting on node i, respectively; Δ i represents the displacement column vector of node i, θ xi , v yi , are the angular displacement in the x direction, the linear displacement in the y direction, and the angular displacement in the z direction of node i, respectively.

[0105] Before introducing the support conditions, K is a known 3n x 3n matrix. F and Δ are unknown column vectors with 3n elements, respectively. Therefore, 3n constraints need to be introduced to solve all the parameters in the original stiffness equation. For a hollow slab girder bridge composed of p slab girders, q groups of transverse beam grids are set, the structure has (p-1)q inter-beam hinge points, 2p nodes at the supports, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joints. For the hinge nodes at the 2p beam end supports, θ xi = 0, v yi = 0, M zi = 0; for the hinge nodes corresponding to the (p-1)q transverse connections, let the nodes on both sides of the hinge be node k and node (k+1), respectively, then M xk = 0, M x(k+1) = 0, F yk + F y(k+1) = 0, M zk = 0, M z(k+1) = 0, v yk = v y(k+1) ; for the nodes at the connections of the pq transverse and longitudinal members, M xi = 0, F yi = FE , M zi = M E , in which F E , M E is the equivalent nodal load of the known bridge deck vehicle load obtained in S1, and the equivalent nodal load distribution method is as follows:

[0106] Since the dimension of the hollow slab beam in the width direction is much smaller than that in the length direction, it is considered that the concentrated force generated by the wheel acts on the longitudinal beam grid, i.e. the eccentric load on a single hollow slab beam is not considered. As shown in Figure 5 , for the longitudinal bar, first, add the additional chain bar and rigid arm to prevent the linear displacement and angular displacement of all nodes, at this time each unit has a fixed end bar end force (hereinafter referred to as fixed end force), and the additional chain bar and rigid arm have additional counterforce and counterforce couple. According to the balance of the node, the values of these additional counterforce and counterforce couple are equal to the algebraic sum of the fixed end forces converging at the node. Then, the additional chain bar and rigid arm are removed, i.e. the above-mentioned additional counterforce and counterforce couple are added to the node as loads after being reversed, these loads are called the equivalent nodal load of the original non-nodal load, and the actual displacement of the node is obtained under the equivalent nodal load. The longitudinal bar under the vehicle load only bears the y-direction concentrated force F, and its fixed end force F F is calculated as follows The equivalent nodal load at any node is the sum of the fixed end forces of all the bars connected thereto,

[0107] The above 3n constraint conditions are combined with the original stiffness equation of the structure to obtain the actual displacement Δ of each node. Finally, the bar end internal force of the bar e is the sum of the fixed end force and the bar end force generated under the action of the equivalent nodal load, F e = F Fe + k e Δ e . After obtaining the bar end bending moment, for the bar without external force, the internal bending moment of the bar is linearly distributed; for the bar with external force, the bar end bending moment is connected and then superimposed with the bending moment diagram of the external force under the simply supported condition to obtain the internal bending moment distribution of the bar. In this way, the bending moment at any length of each hollow slab beam is obtained.

[0108] After obtaining the bending moment at any length of each hollow slab beam, the bending moment envelope area of the region [x i , x i+1 ] on the nth hollow slab at time t is further calculated according to .

[0109] Step 4: Identification of curvature envelope area of region

[0110] According to the rotation angle of the unit node, the curvature envelope area of each stiffness identification region is calculated, and the formula AC (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ), where θ(n,t,x) i Let x be the time t on the nth hollow plate. i At the corner, A C (n,t,x i ) represents the region [x] on the nth hollow plate at time t. i ,x i+1 The curvature envelope area of ​​].

[0111] Step 5: Calculation of regional stiffness

[0112] The uniform stiffness of each test unit is obtained by quotienting the bending moment envelope area and the curvature envelope area of ​​each region and averaging the values ​​over a time scale. The formula is used... The uniform stiffness of the i-th test unit A(n,i) on the n-th hollow plate is calculated, where T is the total effective measurement time.

[0113] Step 6: Calculation of Standardized Stiffness Damage Index

[0114] Based on the uniform stiffness of each test unit, a standardized stiffness damage index is calculated to identify structural damage. The formula is used. The standardized stiffness damage index is calculated to identify damage. Among other things, Let μ be the uniform stiffness of region A(n,i), μ be the mean stiffness of all regions, and σ be the standard deviation of the stiffness of all regions.

[0115] Step 7: Region stiffness identification based on iterative method under damage conditions

[0116] After obtaining the flexural stiffness value under the assumption of structural integrity and identifying the damage, the stiffness value of the intact region is averaged and assigned to the longitudinal beam grid of all intact regions. The bending stiffness value of the damaged area is equal to the previously calculated stiffness value of this area. Torsional stiffness is assumed to change proportionally to bending stiffness.

[0117] Using this stiffness value, update the element stiffness matrix, repeat the matrix displacement method steps, and obtain the bending moment distribution (M(n,t,x)) within the member. i )) loopi Integrating yields the bending moment envelope area (A). M (n,t,x i )) loopi Similarly, using the corner data again, we can calculate the results for each region.

[0118] The rigidity value of the intact area is averaged again and assigned to all intact areas, and the rigidity value of the damaged area is equal to the rigidity value of the area calculated this time. The bending moment calculation and rigidity matrix updating are performed again. Iterative convergence, output area distribution rigidity. Generally, ∈0 takes 0.0001.

[0119] The present application is further described below through specific embodiments, and the test example is the identification of the regional distribution rigidity of a multi-beam system under the action of random vehicle flow.

[0120] Figure 6 A four-lane hollow slab beam bridge is simulated by a model composed of eight aluminum plates representing eight beams, numbered G1 to G8 in turn, each aluminum plate being 4000 mm long, 12 mm thick, and 150 mm wide. A hinge connection is provided every 500 mm along the length direction, and EI = 1.552 x 10 9 N·mm 2 .

[0121] It is considered that the four-lane ranges are G1 and G2, G3 and G4, G5 and G6, and G7 and G8. A corner measuring point is arranged every 500 mm starting from 250 mm from the end of each beam, i.e. eight corner measuring points for each beam, and a total of 64 corner measuring points for the structure. The first measuring point of the first beam is recorded as MP1-1, and the third measuring point of the fifth beam is recorded as MP5-3. A longitudinal region is between two longitudinal corner measuring points. The region between measuring points MP1-1 and MP1-2 is A1-1, and the region between measuring points MP5-3 and MP5-4 is A5-3.

[0122] Due to laboratory site restrictions, two cameras are used to measure the corner angles under the bridge at the same time. During the experiment, two vehicles pass through freely at variable speeds, and the data when both vehicles are passing through the bridge are used for rigidity calculation to avoid the influence of loading and unloading on the experiment.

[0123] The example shows the performance of the method under the conditions of intact structure and damaged structure by damaging part of the structure. Under the condition of multiple damage, region A1-4 is damaged by 15%, and region A8-5 is damaged by 10%.

[0124] In the case of intact structure, a car running experiment is carried out, and the structure load input data and structure output data are collected according to the bridge vehicle load space-time distribution identification method and the corner angle data collection method, as shown in Figs. Figure 7 (a) and 7(c). Using the vehicle load space-time distribution data, the bending moment of the structure at each time and the bending moment envelope area are calculated according to the matrix displacement method. The bending moment envelope area of each region is shown in Fig. Figure 7 (b). At the same time, the curvature envelope area is calculated according to the corner angle data processed by the smoothing filter, as shown in Fig. Figure 7(d) shown. Then, for each region, the data at each time, the uniform stiffness is calculated. Figure 7 (e) respectively show the stiffness calculation results, the normalized stiffness damage index and the stiffness calculation error. It can be seen that when the structure is intact, the normalized stiffness damage index of each region is small, close to 0. The stiffness calculation value is close to the true value, and the maximum error is 2.55%, which is within a reasonable range and meets the engineering requirements.

[0125] The normalized stiffness damage index of each region is calculated as Figure 8 (a) It can be seen that the normalized stiffness damage index of regions A1-4 and A8-5 is significantly greater than that of the remaining regions, and it is identified that regions A1-4 and A8-5 have damage. Note that the normalized stiffness damage index is only used to identify the occurrence of damage, and cannot well characterize the size of the damage. After identifying the damage location, iterative-based damage quantification calculation is performed. The force is distributed using the initial stiffness calculation value, and the stiffness is calculated again; and the newly calculated stiffness is again brought into the force distribution, and the stiffness is iteratively calculated. The relative difference and error of the stiffness of the damaged region and the intact region in 15 iterations are as follows Figure 8 (b) (c). It can be seen that with iteration, the relative difference and error of the stiffness are continuously decreasing. After the 8th iteration, the relative difference of region 2 is less than 0.0001 and continues to decrease thereafter; after the 10th iteration, the relative difference of region 1 is less than 0.0001. Thus, after the 10th iteration, it is considered that the iteration converges, and at this time, the region stiffness calculation value converges near the true value. After iteration to convergence, the stiffness value of each region is obtained as Figure 8 , and the error of the stiffness calculation value relative to the true value is as follows Figure 8 It can be seen that for both undamaged regions and damaged regions, the stiffness identification result close to the true value can be obtained. The maximum error of the stiffness identification result is 2.8%, the error of the damaged region A1-4 is 1.7%, and the error of the damaged region A8-5 is 0.4%.

[0126] The above is only a preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that for ordinary skilled persons in the technical field, some improvements and refinements without departing from the principles of the present application shall be considered as the protection scope of the present application.

Claims

1. A method for identifying the regional distribution of stiffness of a hollow slab bridge based on visual measurement, characterized in that, The bridge unit distributed stiffness identification method is based on the bridge vehicle flow monitoring camera, the online camera under the bridge and the vehicle weight information obtained from the three parties, and the specific steps are as follows: S1, under the action of random traffic load, the vehicle reference point positioning and vehicle information identification are carried out by using the bridge vehicle flow monitoring camera, and the position and size of the vehicle load on the bridge at any time are obtained by combining the vehicle weight data of the toll station; S2, while identifying the bridge load, the rotation angles of the nodes of each test unit in the test bridge area are monitored synchronously by using the online camera under the bridge; S3, according to the position and size of the vehicle load on the bridge at any time in S1 step, the matrix displacement method based on the grillage model is used to calculate the bending moment envelope area of each stiffness identification region at any time; S4, according to the rotation angles of the unit nodes monitored in S2 step, the curvature envelope area of each stiffness identification region is calculated; S5, according to S3 step and S4 step, the quotient of the bending moment envelope area and the curvature envelope area of each region is calculated, and the average value is taken in the time scale to obtain the uniform stiffness of each test unit; S6, according to the uniform stiffness of each test unit, the standardized stiffness damage index is used to identify the structural damage; if the identified structure has no damage, the measurement is ended; if the identified structure has damage, the step S7 is entered; S7, for the damaged structure, the internal force redistribution effect after the regional stiffness damage is considered, and the grillage model parameters are updated by using the iteration method until convergence is obtained, and the regional stiffness under damage is obtained.

2. The visual measurement-based hollow slab bridge regional distributed stiffness identification method according to claim 1, characterized in that: Yolov5s is used to detect vehicles on the whole bridge, vehicle tracking is carried out by using template matching, camera calibration is carried out by using the pixel coordinate information of the feature points on the picture and the known actual space coordinate information, and the homography matrix from the image coordinate system to the bridge world coordinate system is obtained; and the reference point position of each vehicle in the bridge coordinate system at any time is calculated through the homography matrix.

3. The method for identifying the regional distribution of rigidity of a hollow slab bridge based on visual measurement according to claim 1, characterized in that: In S3 step, the bending moment envelope area of each stiffness identification region at any time is calculated by using the matrix displacement method based on the grillage model, including the following steps: S31, the simplified hollow slab bridge is adopted by using the grillage method; the transverse and longitudinal stiffness of the hollow slab is concentrated by using the transverse and longitudinal grillages, and the hinge joints and support joints are hinged; each transverse connection is composed of two rod members, each rod member is hinged to its corresponding longitudinal member, and the two rod members are hinged to each other; S32, for each bar, establish a local coordinate system, set its number in the whole structure as e, two nodes as i, j; take i as the origin, take the direction from i to j as the positive direction of the axis, the vertical upward direction as the axis, establish a right-hand space rectangular coordinate system; axis, establish a right-hand space rectangular coordinate system;​ In each of the bars, the bar end force components are i end x direction bending moment y direction shear force z direction bending moment j end x direction bending moment y direction shear force z direction bending moment According to the displacement method, the stiffness equation of the element is where the end force vector and the end displacement vector are given by Element stiffness matrix is: where (GI t ) e is the torsional stiffness of the bar e, (EI) e is the flexural stiffness of the bar e, l e is the length of the bar e; S33, obtaining the element stiffness matrix After that, the element stiffness matrix k in the global coordinate system is obtained by coordinate transformation using the following equation e : where T is the coordinate transformation matrix, T T denotes the transpose of the coordinate transformation matrix; for the sake of subsequent calculation, k e is divided into where i and j are the numbers of two nodes on the element e; The overall coordinate system and the local coordinate system only rotate around the y-axis, so the coordinate conversion matrix T is: where a is the angle from the x-axis to the line connecting the center of the circle to the point of intersection of the circle with the x-axis. the angle of the axis For the longitudinal grillage, α = 0°; for the transverse grillage, α = 270°; S34, from the unit stiffness matrix k e assembling the global stiffness matrix; the global stiffness matrix K is a 3n x 3n matrix; where the primary sub-block K ii is obtained by superposition of the primary sub-blocks of the respective associated elements of node i, i.e. where the sub-sub-block K im , when i, m are dependent nodes, i.e. the respective sub-sub-block of the cell that links them, i.e. when i, m are independent nodes, i.e. the zero sub-block; S35, after the overall stiffness matrix is combined, the original stiffness equation F = KΔ of the structure is used, Where F is the external force of each node, Δ is the displacement of each node, and each expression is as follows: F = (F1 F2...F n ) T = (M x1 F y1 M z1 M x2 F y2 M z2 ...M xn F yn M zn ) T F i the column vector of external forces acting on node i, M xi ,F yi ,M zi are the external force couple in the x direction, the external force in the y direction, and the external force couple in the z direction, respectively, acting on node i; Δ i the column vector of displacements of node i, are the angular displacement in the x direction, the linear displacement in the y direction, and the angular displacement in the z direction, respectively, of node i; The overall stiffness matrix K is a 3n×3n order matrix, the constraint support condition is introduced, the hollow slab bridge composed of p slab beams is set, q groups of transverse grillages are set, the structure is (p-1)q hinge points between beams, 2p nodes at support joints, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joint; For 2p hinged joints at the supports of the beam ends, let θ xi = 0, v yi = 0, M zi = 0; for (p-1)q hinged joints corresponding to the transverse connections, let the nodes on both sides of the hinged joint be node k and node (k+1) respectively, and the expression is: M xk = 0, M x(k+1) = 0, F yk + F y(k+1) = 0, M zk = 0, M z(k+1) = 0, v yk = v y(k+ 1); For the pq transverse longitudinal bar connection node, its expression is: M xi = 0, F yi = F Eyi , M zi = M Ezi ; F in the above equation Eyi M Ezi The equivalent nodal loads on the nodes to which the known bridge vehicle load in Step S1 is distributed are calculated as follows; The vehicle load on the bridge deck is distributed as equivalent node load. The longitudinal bar under the vehicle load is subjected to a concentrated force F in the y direction, and the fixed-end force F F The calculation formula is as follows: wherein, is the i-end fixed-end shear force, is the j-end fixed-end shear force, is the i-end fixed-end bending moment, is the j-end fixed-end bending moment; F is the y-direction concentrated force acting on the member, l is the length of the member, and a and b are the distances of the concentrated force from i and j, respectively. The equivalent node load on any node i is the sum of the fixed end forces of all rod members connected thereto, and the calculation formula is as follows: where Mx, My, Mz are the equivalent nodal force couples in x, y, z directions respectively acting on node i; Exi , Fx, Fy, Fz are the equivalent nodal forces in x, y, z directions respectively acting on node i; Eyi , Mx, My, Mz are the equivalent nodal force couples in x, y, z directions respectively acting on node i; Ezi , Fx, Fy, Fz are the equivalent nodal forces in x, y, z directions respectively acting on node i; is the fixed-end shear force of member e connected at node i, is the fixed-end bending moment of member e connected at node i, The 3n constraint conditions are combined with the original stiffness equation of the structure to obtain actual displacement Δ of each node. The end force F of the member e is calculated by the following equation e including the x-direction end bending moment the y-direction end shear force the z-direction end bending moment i.e. F e is the sum of the end force due to the fixed end force and the equivalent nodal load where F Ei is the equivalent nodal load of node i, k e is the element stiffness matrix in global coordinate system, Δ e is the element displacement matrix in global coordinate system, (k e Δ e ) i is the i-end bar end force generated under the action of equivalent nodal load; the x-direction bar end bending moment The bending moment distribution inside the bar is obtained by matching the bending moment diagram; that is, the bending moment at any length of each hollow slab beam; S36, after the bending moment of each hollow slab beam at any length is obtained, the bending moment envelope area of the region [x i ,x i+1 ] on the nth hollow slab at time t is further calculated according to the following formula: where A M (n, t, x i ) denotes the bending moment envelope area of the region [x i , x i+1 ] on the n-th hollow slab at time t.

4. The method for identifying regional distribution rigidity of hollow slab bridge based on visual measurement according to claim 1, characterized in that, The curvature envelope area of each stiffness identification region is calculated according to the rotation angle of the unit node in step S4, and the following formula is used for calculation: A C (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ) where θ(n, t, x) is the rotation angle at time t and position x on the nth panel, A(n, t, x) is the curvature envelope area of the region [x, x + Δx] at time t on the nth panel. i i C i i i+1 where θ(n, t, x) is the rotation angle at time t and position x on the nth panel, A(n, t, x) is the curvature envelope area of the region [x, x + Δx] at time t on the nth panel.​​​​​ 5. The method for identifying regional distribution of rigidity of hollow slab bridge based on visual measurement according to claim 1 or 4, characterized in that, Step S5 uses the following formula: wherein, E(n,i) = T / A(n,i) is the uniform stiffness to the i-th test unit A(n,i) on the n-th hollow slab, T is the total length of effective measurement.

6. The method for identifying regional distribution of rigidity of hollow slab bridge based on visual measurement according to claim 1, characterized in that, Step S6 uses the following formula to calculate the normalized stiffness damage index to identify damage; wherein, is the uniform stiffness of region A(n,i), μ is the mean of all region stiffnesses, and σ is the standard deviation of all region stiffnesses.

7. The method for identifying regional distribution of rigidity of hollow slab bridge based on visual measurement according to claim 1, characterized in that, Step S7 updates the beam grid model parameters to obtain the stiffness of the damage area by using the iterative method, and the specific steps are as follows: S71, obtain the bending stiffness value under the assumption that the structure is intact and identify the damage, and average the stiffness values of the intact regions and assign them to all intact regions of the longitudinal beam grid; The expression is as follows: wherein, Ei represents the average value of the uniform rigidity of the intact area under the assumption of intact structure, Ei is the rigidity of the intact area adopted in the first iteration; The bending stiffness value of the damage area is equal to the stiffness value of this area under the assumption that the structure is intact, wherein E(n,i) denotes the uniform stiffness of the damaged area (n,i) under the perfect structure assumption, E(n,i) is the stiffness of the damaged area (n,i) used for the first iteration; For torsional stiffness, it is considered to change in proportion to the bending stiffness; S72, update the unit stiffness matrix k with the stiffness value obtained in S71 e , repeat steps S34-S36 to obtain the bending moment envelope area (A M (n, t, x i )) loop1 ; obtain the uniform stiffness of the region (n, i) of the first iteration for each region using the curvature envelope area A C (n, t, x i ) S73, again average the stiffness values of the intact regions and assign them to all intact regions, wherein, is the uniform stiffness average value of the intact zone obtained for the (i-1)th iteration, is the stiffness value of the intact zone used for the ith iteration; The bending stiffness value of the damage area is equal to the stiffness value of this area calculated last time, wherein E(n,i) is the homogenous stiffness of the damaged area (n,i) obtained for the (i-1)th iteration, E(n,i) is the homogenous stiffness of the damaged area (n,i) adopted for the ith iteration; For torsional stiffness, it is considered to change in proportion to the bending stiffness; S74, update the unit stiffness matrix k with the stiffness value obtained in S73 e , repeat steps S34-S36 to obtain the bending moment envelope area (A M (n, i) of the i-th iteration i ) loopi ; calculate the uniform stiffness of the region (n, i) for each region using the curvature envelope area A C (n, i) i ​ S75, repeating steps S73-S74 until convergence, and where ∈ is the relative difference; in order to take into account the calculation efficiency and accuracy, ∈0 is taken as 0.0001; at this time, is the final uniform stiffness calculation result of the region (n, i).

Images