Vision measurement-based hollow slab bridge regional distribution rigidity identification method

Abstract

The invention belongs to the technical field of building structure health monitoring, and particularly relates to a hollow slab bridge regional distribution rigidity identification method based on vision measurement, which can realize regional rigidity identification of a hollow slab bridge. The method comprises the following specific steps: under the action of random traffic load, carrying out vehicle reference point positioning and vehicle information identification by using a traffic flow monitoring camera on a bridge, and obtaining the position and size of the vehicle load on the bridge floor at any moment by combining the vehicle weight data of a toll station; when the load of the bridge floor is identified, the corner of each test unit node in the test bridge area is synchronously monitored by using an under-bridge online camera; according to the position and size of the bridge deck vehicle load at any moment, calculating the bending moment envelope area of each rigidity identification area at any moment by utilizing a matrix displacement method based on a beam lattice model; calculating the curvature envelope area of each rigidity identification area according to the monitored rotation angle of the unit node; performing quotient on the bending moment envelope area and the curvature envelope area of each region, and taking a mean value at a time scale to obtain the uniform rigidity of each test unit; according to the uniform rigidity of each test unit, calculating a standardized rigidity damage index to identify structural damage; if the identified structure is not damaged, ending the measurement; and if the identified structure is damaged, considering the internal force redistribution effect after the regional rigidity is damaged for the damaged structure, and updating beam lattice model parameters by using an iteration method until convergence to obtain the regional rigidity under the damage. According to the method, large width span ratio and obvious transverse distribution effect of the hollow slab bridge are considered, structure bending moment distribution is calculated by utilizing a matrix displacement method based on a beam lattice model, and the method is rapid and convenient, high in testing precision and suitable for rapid testing of regional distribution rigidity of the hollow slab bridge.

Description

Technical Field

[0001] The present invention belongs to the technical field of building structure health monitoring, and in particular relates to a method for identifying regional distribution stiffness of a hollow slab bridge based on visual measurement, which can realize regional stiffness identification of the hollow slab bridge. Background Art

[0002] As an important part of the modern transportation network, bridges may suffer fatigue damage or even structural damage prematurely under the long-term unreasonable vehicle action. Using structural monitoring technology to timely detect structural diseases and hidden dangers is expected to solve the problem of structural service safety and provide strong support for achieving long-term structural life. In the design and construction of small and medium-span bridges in my country, prefabricated reinforced concrete beam bridges are widely used due to their simple structure and good mechanical properties. At the same time, hollow slabs have unique advantages such as small self-weight, simple prefabrication, and easy construction. Therefore, in the construction of small and medium-span bridges, prefabricated hollow slab beam bridges have become the first choice, accounting for more than 50% of small and medium-span beam bridges in my country. Affected by factors such as design, construction, natural erosion and operating environment, prestressed concrete hollow slab beam bridges will gradually suffer from material degradation, cracks, structural damage and other diseases as their service time increases.

[0003] In the field of structural identification, if the input and output information of the structure can be effectively utilized, it can provide strong support for the identification, evaluation and maintenance of bridge structures. Based on the non-contact measurement advantages of visual technology, it is possible to monitor the spatiotemporal distribution of vehicle weight on the bridge deck to obtain structural load input information while obtaining real-time structural response output measurement based on online cameras.

[0004] Accurately identifying the essential characteristics of bridges can provide important structural health information, which helps to timely discover potential material degradation and structural crack disease problems and take corresponding maintenance and reinforcement measures to ensure the safe operation of bridges. Existing health monitoring research has made some progress in the perception of multi-dimensional visual information such as the acquisition of spatiotemporal distribution of bridge traffic based on visual technology (input) and structural response monitoring based on optical measurement (output), but there are still limitations in using health monitoring data to obtain structural essential characteristics information and evaluate the service status of hollow slab girder bridges.

[0005] Environmental vibration tests that rely only on "output" information can only identify basic modal parameters, which are weakly correlated with structural performance and cannot effectively support structural performance evaluation; impact vibration tests with known "input-output" information can identify deep-level parameters such as the structural flexibility matrix, but their characteristics of requiring input of impact load measurement limit their use to traffic closure, which brings inconvenience to the inspection and monitoring of large-scale and wide-ranging small and medium-sized bridges; existing mobile load tests based on sports car tests use quasi-static influence line indicators to reflect the overall force performance of the structure, but they also have the limitation of closed traffic. Using the vehicle load data on the bridge as the structural input and the structural deformation data as the structural response output, there is no need to close traffic, and deep-level structural parameters can be obtained. Therefore, the development of a hollow slab bridge damage identification method based on vehicle flow identification and structural displacement has practical engineering significance and clear engineering needs. Summary of the invention

[0006] Technical problem to be solved: The present invention discloses a method for identifying the regional distribution stiffness of hollow slab bridges based on visual measurement, which has the advantages of fast, convenient and non-contact measurement, overcomes the problem of the existing method requiring traffic closure, and has a good prospect of being widely used in the identification of regional stiffness of hollow slab bridges.

[0007] The present invention adopts the following technical solution:

[0008] A method for identifying regional distributed stiffness of a hollow slab bridge based on visual measurement comprises the following steps:

[0009] S1, under the action of random traffic load, the traffic monitoring camera on the bridge is used to locate the vehicle reference point and identify the vehicle information, and combined with the vehicle weight data of the toll station, the position and size of the vehicle load on the bridge deck at any time are obtained;

[0010] S2, while identifying the bridge deck load, uses an online camera to synchronously monitor the rotation angles of each test unit node in the test bridge area.

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

[0012] S4, calculating the curvature envelope area of ​​each stiffness identification region according to the rotation angle of the unit node.

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

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

[0015] S7, for damaged structures, the internal force redistribution effect after regional stiffness damage is considered, and the beam grillage model parameters are updated by iterative method until convergence to obtain the regional stiffness under damage.

[0016] Furthermore, in step S1, the vehicle reference point positioning and vehicle information recognition are performed using the traffic monitoring camera on the bridge, and the position and size of the vehicle load on the bridge deck at any time are obtained in combination with the vehicle weight data of the toll station, which includes the following sub-steps:

[0017] S11, using Yolov5s to detect vehicles on the bridge image, for each frame of the picture, using the identified vehicle mask to identify the lower left point of the vehicle head (specifically: the leftmost point among the three lowest feature points on the mask contour) as a reference point;

[0018] S12, using template matching to track vehicles, calculate the correlation coefficient between each vehicle sub-image in the next frame and each vehicle sub-image in the previous frame to form a correlation matrix, find the peak value of each row from the correlation matrix, and the row and column where the peak value is located represent the same vehicle. When each vehicle first enters the field of view, it is given a label, and the same label is given to the vehicle matching this vehicle in the next frame. Through the cycle, the label of this vehicle is the same in all frames, so that vehicle tracking and reference point matching are achieved;

[0019] S13, using the pixel coordinate information of the feature points on the image and the known actual space coordinate information to perform camera calibration, and obtain a homography matrix from the image coordinate system to the bridge deck world coordinate system; and calculating the reference point position of each vehicle in the bridge deck coordinate system at any time through the homography matrix;

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

[0021] S15, at the same time, YOLOv5s is used to identify the border of the license plate; LPRNet is used to further identify the license plate characters, thereby corresponding to the vehicle weight data.

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

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

[0024] Furthermore, in step S3, according to the size and position of the traffic load, the moment envelope area of ​​each stiffness identification area at any time is calculated using a matrix displacement method based on a beam grid model, including the following steps:

[0025] S31, the beam grid method is used to simplify the hollow slab beam bridge. The transverse and longitudinal beam grids are used to concentrate the transverse and longitudinal stiffness of the hollow slab, and the hinge joints and supports are hinged. Each transverse connection consists of two rods, each of which is rigidly connected to its corresponding longitudinal member, and the two rods are hinged to each other.

[0026] S32, for each rod, establish a local coordinate system, set its number in the whole structure as e, it 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, vertically upward direction is Axis, establish a right-handed rectangular coordinate system. In each rod, ignore the effect of axial force; assume that the node only moves in 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 in the x direction at the i end Shear force in the y direction Bending moment in z direction and the bending moment in the x direction at end j Shear force in the y direction Bending moment in z direction According to the displacement method, the stiffness equation of the element is The rod end force vector and the rod end displacement vector respectively Element stiffness matrix for

[0027]

[0028] Among them (GI t ) e is the torsional stiffness of member e, (EI) e is the bending stiffness of member e, l e is the length of the rod e. In this method, since the matrix displacement method is only used to calculate the distribution of force without outputting the structural deformation, the material elastic modulus can take any value. Section bending moment of inertia I = dh 3 / 12, torsional moment of inertia I t =0.33dh 3 For the transverse member, since the hinge joint does not transmit torque, the torsional stiffness of the transverse member does not affect the structural response. For convenience, the transverse member I t = 0, since the plate beam will produce elastic deflection in the longitudinal direction, but can be considered rigid in the transverse direction, that is, the bending stiffness of the transverse rod in the model is much greater than that of the longitudinal rod, and can be considered infinite. Here, the bending moment of inertia of the section is assigned I = lh 3 / 36, to adapt to the calculation of matrix displacement method.

[0029] S33, after obtaining the unit stiffness matrix, use Perform coordinate transformation to obtain the unit stiffness matrix in the global coordinate system and divide it into blocks for 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 distance from the x-axis to The angle between the longitudinal beam grid and the transverse beam grid is α = 0°; for the transverse beam grid, α = 270°.

[0032] S34, assemble the overall stiffness matrix from the unit stiffness matrix. Assume that there are n nodes in the beam grid method model, then the overall stiffness matrix K has n 2 sub-blocks, each of which is a 3×3 square matrix, so K is a 3n×3n square matrix. ii is obtained by superimposing the main sub-blocks of each relevant unit of node i, that is, The sub-block K im ,When i,m are related nodes, they are the corresponding sub-blocks of the unit connecting them, i.e. When i and m are non-related nodes, they are zero sub-blocks.

[0033] S35, after being combined into the overall 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.

[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 Represents the external force column vector of node i, M xi ,F yi ,M zi are the external couple along the x direction, the external force along the y direction, and the external couple along 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 of node i along the x direction, the linear displacement along the y direction, and the angular displacement along the z direction, respectively.

[0037] Before the support condition is introduced, K is a 3n×3n known square matrix. F and Δ are unknown column vectors with 3n elements. Therefore, 3n constraints need to be introduced to solve all the parameters in the original stiffness equation. For a hollow slab girder bridge consisting of p slab beams, q groups of transverse beam grids are set, and the structure has (p-1)q hinge points between beams, 2p nodes at the supports, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joint. For the 2p hinge nodes at the beam end supports, there is θ xi =0, v yi =0,M zi = 0; for the hinges 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 where pq transverse and longitudinal members are connected, M xi =0, F yi =F E , M zi =M E , where F E ,M E The equivalent node loads on the nodes are allocated to the known bridge deck vehicle loads obtained in S1. The equivalent node load allocation method is as follows:

[0038] Since the width dimension of the hollow slab beam is much smaller than the length dimension, it is assumed that the concentrated forces generated by the wheels all act on the longitudinal beam grid, that is, the eccentric load of the concentrated forces on a single hollow slab beam is not considered. For the longitudinal members, first, additional links and rigid arms are added to prevent the linear and angular displacements of all nodes. At this time, each unit has a fixed end rod end force (hereinafter referred to as the fixed end force), and the additional links and rigid arms have additional reactions and reaction couples. From the node equilibrium, it can be seen that the values ​​of these additional reactions and reaction couples are equal to the algebraic sum of the fixed end forces converging at the node. Then, the additional links and rigid arms are cancelled, that is, the above-mentioned additional reactions and reaction couples are reversed and added to the nodes as loads. These loads are called equivalent node loads of the original non-node loads. The actual displacement of the node is obtained under the equivalent node load. Under the vehicle load, the longitudinal member is only subjected to the concentrated force F in the y direction, and its fixed end force F F The calculation is as follows The equivalent node load on any node is the sum of the fixed end forces of all the members connected to it.

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

[0040] The rod end internal force F of member e is calculated by the following formula e , including the bending moment at the rod end in the x direction Shear force at the rod end in the y direction Bending moment at the rod end in z direction That is F e It is the sum of the rod end force generated by the fixed end force and the equivalent node load:

[0041]

[0042] where F Ei is the equivalent node load of node i, k e is the element stiffness matrix in the global coordinate system, Δ e is the unit displacement matrix in the global coordinate system, (k e Δ e ) i The rod end force generated by the equivalent node load is Combined with the bending moment diagram, the bending moment distribution inside the member is obtained; that is, 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 area [x i ,x i+1 ]’s bending moment envelope area.

[0043] Furthermore, in step S4, the curvature envelope area of ​​each stiffness identification area is calculated according to the rotation angle of the unit node, using formula A C (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ), where θ(n,t,x i ) is the x value of the nth hollow plate at time t i The corner at A C (n,t,x i ) is the area [x i ,x i+1 ] is the curvature envelope area.

[0044] Furthermore, step S5 adopts the formula 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.

[0045] Furthermore, step S6 adopts the formula The normalized stiffness damage index is calculated to identify damage. is the uniform stiffness of region A(n,i), μ is the mean stiffness of all regions, and σ is the standard deviation of the stiffness of all regions.

[0046] Furthermore, step S7 uses an iterative method to update the beam grillage model parameters to obtain the stiffness of the damaged area. The specific steps are:

[0047] S71, after obtaining the bending stiffness value under the assumption that the structure is intact and identifying the damage, the stiffness value of the intact area is averaged and assigned to all the longitudinal beam grids in the intact area; the expression is as follows:

[0048]

[0049] in, represents the average uniform stiffness of the intact area under the assumption that the structure is intact, is the intact region stiffness adopted for the first iteration.

[0050] The bending stiffness value of the damaged area is equal to the stiffness value of this area under the intact assumption.

[0051]

[0052] in, represents the uniform stiffness of the damaged area (n,i) under the assumption that the structure is intact, is the stiffness of the damaged region (n,i) adopted in the first iteration;

[0053] For torsional stiffness, it is considered to vary in proportion to bending stiffness;

[0054] S72, using the stiffness value obtained in S71, update the unit stiffness matrix k e , repeat steps S34 to S36 to obtain the moment envelope area (A) obtained in the first iteration M (n,t,x i )) loop 1; Using the curvature envelope area A C (n,t,x i ), for each region, find the uniform stiffness of the region (n,i) obtained in the first iteration

[0055]

[0056] S73, again averaging the stiffness values ​​of the intact regions and assigning the average stiffness values ​​to all intact regions.

[0057]

[0058] in, is the average uniform stiffness of the intact area obtained in the (i-1)th iteration, is the intact region stiffness value adopted in the i-th iteration.

[0059] The bending stiffness value of the damaged area is equal to the stiffness value of this area calculated last time.

[0060]

[0061] in, is the uniform stiffness of the damaged area (n,i) obtained in the (i-1)th iteration, is the uniform stiffness of the damaged region (n,i) adopted in the i-th iteration.

[0062] The present invention provides a method for identifying the regional distribution stiffness of a hollow slab bridge based on visual measurement. First, a computer vision method is used to identify the vehicle position and information, and the spatiotemporal distribution of the bridge deck vehicle weight is obtained in combination with the vehicle weight data of the toll station. Then, the matrix displacement method is used to calculate the moment envelope area of ​​each region; at the same time, optical measurement is used to realize the synchronous measurement of the structural angular response, and the curvature envelope area is calculated accordingly. The stiffness of each test area can be obtained by taking the average of the quotient of the moment envelope area and the curvature envelope area on a time scale. Thereafter, the standardized stiffness damage index is calculated to identify the damage. For damaged structures, the internal force redistribution effect is considered, and the model parameters are updated using an iterative method to accurately quantify the regional stiffness under damage.

[0063] Beneficial effects:

[0064] First, the method for identifying the regional distribution stiffness of hollow slab bridges based on visual measurement of the present invention solves the shortcomings of traditional bridge multi-point stiffness measurement, which requires a large number of contact sensors, requires traffic closure, and has high testing costs.

[0065] Second, this method takes into account the large width-to-span ratio and obvious lateral distribution effect of hollow slab bridges. It uses the matrix displacement method based on the beam grid model to calculate the structural bending moment distribution. It is fast, convenient and has high test accuracy. It is suitable for the rapid test of regional distribution stiffness of hollow slab bridges.

[0066] Third, this method considers the internal force redistribution effect of the regional stiffness of the hollow slab beam bridge after damage, and uses the iterative method to update the beam grid model parameters, which can accurately quantify the regional stiffness under damage and is suitable for bridge damage status assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic diagram of the principle of a method for identifying the regional distribution stiffness of a hollow slab bridge based on visual measurement according to an embodiment of the present invention;

[0068] Figure 2 This is a schematic diagram of vehicle tracking based on computer vision;

[0069] Figure 3 This is a schematic diagram of vehicle information recognition based on computer vision;

[0070] Figure 4 It is a schematic diagram of the beam grid method principle;

[0071] Figure 5 It is a schematic diagram of the equivalent load distribution principle;

[0072] Figure 6 Provide a layout diagram for the laboratory;

[0073] Figure 7 The stiffness calculation diagram for the intact working condition is shown;

[0074] Figure 8 This is a diagram showing the stiffness calculation for the damage condition. DETAILED DESCRIPTION

[0075] The following examples will enable those skilled in the art to more fully understand the present invention, but are not intended to limit the present invention in any way.

[0076] See also Figure 1 This embodiment discloses a bridge distributed stiffness identification method based on curvature envelope area and microwave radar, and the bridge unit distributed stiffness identification method comprises the following steps:

[0077] S1, under the action of random traffic load, the traffic monitoring camera on the bridge is used to locate the vehicle reference point and identify the vehicle information, and combined with the vehicle weight data of the toll station, the position and size of the vehicle load on the bridge deck at any time are obtained;

[0078] S2, while identifying the bridge deck load, uses an online camera to synchronously monitor the rotation angles of each test unit node in the test bridge area.

[0079] S3, according to the size and location of the traffic load, the matrix displacement method based on the beam grid model is used to calculate the moment envelope area of ​​each stiffness identification area at any time.

[0080] S4, calculating the curvature envelope area of ​​each stiffness identification region according to the rotation angle of the unit node.

[0081] S5, the quotient of the moment envelope area and the curvature envelope area of ​​each region is taken, and the average is taken on the time scale to obtain the uniform stiffness of each test unit.

[0082] S6, according to the uniform stiffness of each test unit, the normalized stiffness damage index is calculated to identify structural damage.

[0083] S7, for damaged structures, the internal force redistribution effect after regional stiffness damage is considered, and the beam grillage model parameters are updated by iterative method until convergence to obtain the regional stiffness under damage.

[0084] The specific steps include:

[0085] Step 1: Obtain the location and size of the bridge deck vehicle load

[0086] The traffic monitoring camera is installed above the bridge deck, and the direction of the camera is adjusted so that it completely covers the bridge deck. The data acquisition system collects traffic images and saves the collected image sequence files.

[0087] like Figure 2As shown, Yolov5s is used to detect vehicles on the entire bridge. For each frame of the picture, the identified vehicle mask is used to identify the lower left point of the front of the vehicle (specifically: the leftmost point of the three lowest feature points on the mask contour) as a reference point; template matching is used to track the vehicle, and the correlation coefficient is calculated between each vehicle sub-image in the next frame and each vehicle sub-image in the previous frame to form a correlation matrix. The peak value of each row is found from the correlation matrix, and the rows and columns where the peak value is located represent the same vehicle. Each vehicle is assigned a label when it first enters the field of view, and the same label is assigned to the vehicle that matches this vehicle in the next frame. Through the loop, the vehicle has the same label 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; and the reference point position of each vehicle in the bridge deck coordinate system at any time is calculated through the homography matrix. As shown Figure 3 As shown in the figure, when a vehicle enters the vehicle information recognition area, Yolov5s is used to perform mask recognition of the vehicle and wheels, and the lowest point of the mask contour is taken as the reference point of the vehicle and wheels; the position of the wheel and vehicle reference points in the bridge deck coordinate system is further calculated based on the homography matrix. At the same time, YOLOv5s is used to recognize the border of the license plate; LPRNet is used to further recognize the license plate characters, thereby corresponding to the vehicle weight data. The load position is obtained by combining the vehicle reference point position and the wheel relative to the reference point; the load size is obtained by combining the vehicle weight data.

[0088] Step 2: Node corner synchronization monitoring

[0089] An online camera is set up under the bridge, and the direction and focal length of the online camera are adjusted to make all measuring points have clear images in the field of view. The data acquisition system collects corner monitoring images and saves the collected image sequence files, which are aligned in time with the image sequence collected by the traffic monitoring camera on the bridge according to the timestamp.

[0090] When processing images, the array target image sequence with multiple feature point position constraints is used as the visual input, and the geometric relationship between the camera and the moving target coordinate system is constructed to calculate the bridge turning angle. An adaptive multi-wave gate tracking based on the centroid array and coordinate system transformation matrix purification optimization are used to improve the robustness of turning angle monitoring.

[0091] Step 3: Identification of the Moment Envelope Area of ​​the Region

[0092] like Figure 4The beam grid method is used to simplify the hollow slab beam bridge. The transverse and longitudinal beam grids are used to concentrate the transverse and longitudinal stiffness of the hollow slab, and the hinge joints and supports are hinged. Each transverse connection consists of two rods, each of which is rigidly connected to its corresponding longitudinal member, and the two rods are hinged to each other. For each rod, a local coordinate system is established, and its number in the entire structure is set as 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, vertically upward direction is Axis, establish a right-handed rectangular coordinate system. In each rod, ignore the effect of axial force; assume that the node only moves in 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 in the x direction at the i end Shear force in the y direction Bending moment in z direction and the bending moment in the x direction at end j Shear force in the y direction Bending moment in z direction

[0093] According to the displacement method, the stiffness equation of the element is The rod end force vector and the rod end displacement vector respectively Element stiffness matrix for:

[0094]

[0095] Among them (GI t ) e is the torsional stiffness of member e, (EI) e is the bending stiffness of member e, l e is the length of the rod e. In this method, since the matrix displacement method is only used to calculate the distribution of force without outputting the structural deformation, the material elastic modulus can take any value. Section bending moment of inertia I = dh 3 / 12, torsional moment of inertia I t =0.33dh 3 For the transverse member, since the hinge joint does not transmit torque, the torsional stiffness of the transverse member does not affect the structural response. For convenience, the transverse member I t = 0, since the plate beam will produce elastic deflection in the longitudinal direction, but can be considered rigid in the transverse direction, that is, the bending stiffness of the transverse rod in the model is much greater than that of the longitudinal rod, and can be considered infinite. Here, the bending moment of inertia of the section is assigned I = lh 3 / 36, to adapt to the calculation of matrix displacement method.

[0096] After obtaining the element stiffness matrix, use Perform coordinate transformation to obtain the unit stiffness matrix in the global coordinate system and divide it into blocks for 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

[0097]

[0098] Where α is the distance from the x-axis to The angle between the longitudinal beam grid and the transverse beam grid is α = 0°; for the transverse beam grid, α = 270°.

[0099] The overall stiffness matrix is ​​assembled from the unit stiffness matrix. Assuming there are n nodes in the beam grid method model, the overall stiffness matrix K has n 2 sub-blocks, each of which is a 3×3 square matrix, so K is a 3n×3n square matrix. ii is obtained by superimposing the main sub-blocks of each relevant unit of node i, that is, The sub-block K im ,When i,m are related nodes, they are the corresponding sub-blocks of the unit connecting them, i.e. When i and m are non-related nodes, they are zero sub-blocks.

[0100] After combining into the overall 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.

[0101] 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

[0102]

[0103] Here, F i Represents the external force column vector of node i, M xi ,F yi ,M zi are the external couple along the x direction, the external force along the y direction, and the external couple along 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 of node i along the x direction, the linear displacement along the y direction, and the angular displacement along the z direction, respectively.

[0104] Before the support condition is introduced, K is a 3n×3n known square matrix. F and Δ are unknown column vectors with 3n elements. Therefore, 3n constraints need to be introduced to solve all the parameters in the original stiffness equation. For a hollow slab girder bridge consisting of p slab beams, q groups of transverse beam grids are set, and the structure has (p-1)q hinge points between beams, 2p nodes at the supports, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joint. For the 2p hinge nodes at the beam end supports, there is θ xi =0, v yi =0,M zi = 0; for the hinges 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 where pq transverse and longitudinal members are connected, M xi =0, F yi =F E , M zi =M E , where F E ,M E The equivalent node loads on the nodes are allocated to the known bridge deck vehicle loads obtained in S1. The equivalent node load allocation method is as follows:

[0105] Since the width of the hollow slab beam is much smaller than its length, it is assumed that the concentrated force generated by the wheels acts on the longitudinal beam grid, that is, the eccentric load of the concentrated force on a single hollow slab beam is not considered. Figure 5 As shown, for the longitudinal members, first, additional links and rigid arms are added to prevent the linear and angular displacements of all nodes. At this time, each unit has a fixed end rod end force (hereinafter referred to as the fixed end force), and the additional links and rigid arms have additional reactions and reaction couples. From the node equilibrium, it can be seen that the values ​​of these additional reactions and reaction couples are equal to the algebraic sum of the fixed end forces converging at the node. Then, the additional links and rigid arms are cancelled, that is, the above-mentioned additional reactions and reaction couples are reversed and added to the nodes as loads. These loads are called equivalent node loads of the original non-node loads. The actual displacement of the node is obtained under the equivalent node load. Under the vehicle load, the longitudinal member is only subjected to the concentrated force F in the y direction, and its fixed end force F F The calculation is as follows The equivalent node load on any node is the sum of the fixed end forces of all the members connected to it.

[0106] The above 3n constraints are combined with the original stiffness equation of the structure to solve the actual displacement Δ of each node. Finally, the rod end internal force of member e is the sum of the fixed end force and the rod end force generated by the equivalent node load, F e =F Fe +k e Δ e After obtaining the rod end bending moment, for the rod without external force, the bending moment inside the rod is linearly distributed; for the rod with external force, the segment superposition method is used to connect the rod end bending moments and then superimpose the bending moment diagram of the external force under the simply supported condition to obtain the bending moment distribution inside the rod. In this way, the bending moment at any length of each hollow slab beam is obtained.

[0107] After obtaining the bending moment at any length of each hollow slab beam, further Calculate the area [x i ,x i+1 ]’s bending moment envelope area.

[0108] Step 4: Identification of the curvature envelope area of ​​the region

[0109] According to the rotation angle of the unit node, the curvature envelope area of ​​each stiffness identification area is calculated using the formula A C (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ), where θ(n,t,x i ) is the x value of the nth hollow plate at time t i The corner at A C (n,t,x i ) is the area [x i ,x i+1 ] is the curvature envelope area.

[0110] Step 5: Area Stiffness Calculation

[0111] The quotient of the moment envelope area and the curvature envelope area in each region is calculated and the average is taken on a time scale to obtain the uniform stiffness of each test unit. 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.

[0112] Step 6: Normalized Stiffness Damage Index Calculation

[0113] According to the uniform stiffness of each test unit, the standardized stiffness damage index is calculated to identify structural damage. The formula is used The normalized stiffness damage index is calculated to identify damage. is the uniform stiffness of region A(n,i), μ is the mean stiffness of all regions, and σ is the standard deviation of the stiffness of all regions.

[0114] Step 7: Iterative method based identification of regional stiffness in the damage case

[0115] After obtaining the bending stiffness value under the assumption that the structure is intact and identifying the damage, the stiffness values ​​of the intact area are averaged and assigned to the longitudinal beam grids of all intact areas. The bending stiffness value of the damaged area is equal to the stiffness value of this area calculated last time. As for the torsional rigidity, it is considered that it varies in proportion to the bending rigidity.

[0116] Using this stiffness value, the element stiffness matrix is ​​updated and the matrix displacement method steps are repeated to obtain the moment distribution (M(n, t, x i )) loop i , integrate to get the moment envelope area (A M (n,t,x i )) loop i Similarly, using the angle data again, for each area, find

[0117] The stiffness values ​​of the intact areas are averaged again and assigned to all intact areas. The stiffness value of the damaged area is equal to the stiffness value of this area calculated this time. The bending moment calculation and stiffness matrix update are performed again. After iterative convergence, the regional distribution stiffness is output. Generally, ∈0 is set to 0.0001.

[0118] The present invention is further described below by means of a specific embodiment, wherein a test example is the identification of regional distribution stiffness of a multi-beam system under the action of random traffic flow.

[0119] Figure 6 The middle model simulates a four-lane hollow slab bridge. The model consists of eight aluminum plates, representing eight beams, numbered G1 to G8. Each aluminum plate is 4000 mm long, 12 mm thick, and 150 mm wide. A hinge connection is set every 500 mm along the length direction. EI = 1.552×10 9 N·mm 2 .

[0120] The four lanes are considered to be G1 and G2, G3 and G4, G5 and G6, G7 and G8. A corner measuring point is arranged every 500mm starting from 250mm from the end of each beam, that is, 8 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. There is a longitudinal area between two longitudinal corner measuring points, the area between measuring points MP1-1 and MP1-2 is A1-1, and the area between measuring points MP5-3 and MP5-4 is A5-3.

[0121] For example, due to the limitation of laboratory space, two cameras were used under the bridge to measure the turning angle simultaneously. During the experiment, two vehicles passed through the bridge at different speeds, and the stiffness was calculated based on the data of both vehicles passing through the bridge to avoid the influence of loading and unloading on the experiment.

[0122] The embodiment shows the performance of the method under the condition of intact structure and damaged structure by destroying part of the structure. Under the condition of multiple damage, the area A1-4 is damaged by 15%; the area A8-5 is damaged by 10%.

[0123] When the structure is intact, a sports car experiment is carried out. The structural load input data and structural output data are collected according to the bridge vehicle load time-space distribution identification method and the angle data collection method. Figure 7 (a) and 7(c). Using the on-board spatial and temporal distribution data, the bending moment and the moment envelope area of ​​the structure at each moment are calculated according to the matrix displacement method. The moment envelope area of ​​each region is shown in Figure 7 (b) At the same time, the curvature envelope area is calculated based on the rotation angle data processed by the smoothing filter, as shown in Figure 7 (d) is shown. Then, for the data of each region at each time, the uniform stiffness is calculated. Figure 7 (e) shows the stiffness calculation results, the standardized stiffness damage index, and the stiffness calculation error. It can be seen that when the structure is intact, the standardized stiffness damage index of each region is very small, close to 0. The stiffness calculation values ​​are close to the true value, with a maximum error of 2.55%, which is within a reasonable range and meets the engineering requirements.

[0124] The damage of area A1-4 is 15%, and that of area A8-5 is 10%. The standardized stiffness damage index of each area is calculated as follows: Figure 8(a). It can be clearly seen that the normalized stiffness damage index of regions A1-4 and A8-5 is significantly greater than that of the rest of the regions, and it is identified that damage has occurred in regions A1-4 and A8-5. Note that the normalized stiffness damage index is only used to identify the occurrence of damage, and cannot well characterize the magnitude of the damage. After the damage location is identified, an iteration-based quantitative damage calculation is performed. The initial stiffness calculation value is used to distribute the force, and the stiffness is calculated again; the newly calculated stiffness is brought back into the force distribution, and the stiffness is calculated iteratively. The relative difference in stiffness and the decrease in error between the damaged area and the intact area in 15 iterations are shown in the figure below. Figure 8 (b)(c). It can be seen that with the iteration, the relative difference and error of stiffness are constantly 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. Therefore, after the 10th iteration, the iteration is considered to have converged. At this time, the calculated value of regional stiffness converges to the true value. After iterating to convergence, the stiffness values ​​of each region are obtained as follows Figure 8 , the error of the calculated stiffness value relative to the true value is as follows Figure 8 It can be seen that both the undamaged area and the damaged area can obtain stiffness identification results close to the true value. The maximum error of the stiffness identification result is 2.8%, the error of the damaged area A1-4 is 1.7%, and the error of the damaged area A8-5 is 0.4%.

[0125] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement, characterized in that: The bridge unit distributed stiffness identification method is based on the traffic monitoring camera on the bridge, the online camera under the bridge and the vehicle weight information obtained from three parties. The specific steps are as follows: S1, under the action of random traffic load, the traffic monitoring camera on the bridge is used to locate the vehicle reference point and identify the vehicle information, and combined with the vehicle weight data of the toll station, the position and size of the vehicle load on the bridge deck at any time are obtained; S2, while identifying the bridge deck load, uses the online camera under the bridge to synchronously monitor the rotation angles of each test unit node in the test bridge area; S3, according to the position and size of the bridge deck vehicle load at any time in step S1, the moment envelope area of ​​each stiffness identification area at any time is calculated using a matrix displacement method based on a beam grid model; S4, calculating the curvature envelope area of ​​each stiffness identification region according to the rotation angle of the unit node monitored in step S2; S5, according to step S3 and step S4, the quotient of the moment envelope area and the curvature envelope area of ​​each region is calculated, and the average is taken on a 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 calculated to identify the structural damage; if the identified structure is not damaged, the measurement is terminated; if the identified structure is damaged, the step S7 is entered; S7, for damaged structures, the internal force redistribution effect after regional stiffness damage is considered, and the beam grillage model parameters are updated by iterative method until convergence to obtain the regional stiffness under damage.

2. The method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement according to claim 1 is characterized in that: Yolov5s is used to detect vehicles on the entire bridge, template matching is used for vehicle tracking, and the camera is calibrated using the pixel coordinate information of the feature points on the image and the known actual space coordinate information 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 moment is calculated through the homography matrix.

3. The method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement according to claim 1 is characterized in that: In step S3, the matrix displacement method based on the beam grillage model is used to calculate the moment envelope area of ​​each stiffness identification area at any time, including the following steps: S31. Use the beam grid method to simplify the hollow slab beam bridge; use the transverse and longitudinal beam grids to concentrate the transverse and longitudinal stiffness of the hollow slab respectively, and hinge at the hinge joint and support; each transverse connection is composed of two rods, each rod is rigidly connected to its corresponding longitudinal member, and the two rods are hinged to each other; S32. For each rod, establish a local coordinate system, set its number in the whole structure as e, and the two nodes as i and j; take i as the origin and the direction from i to j as The positive direction of the axis, vertically upward direction is Axis, establish a right-handed rectangular coordinate system; In each member, the end force component has a bending moment in the x direction at end i Shear force in the y direction Bending moment in z direction Bending moment in x direction at end j Shear force in the y direction Bending moment in z direction According to the displacement method, the stiffness equation of the element is The rod end force vector and the rod end displacement vector They are as follows: Element stiffness matrix for: Among them (GI t ) e is the torsional stiffness of member e, (EI) e is the bending stiffness of member e, l e is the length of rod e; S33, get the unit stiffness matrix After that, the coordinate transformation is performed using the following formula to obtain the unit stiffness matrix k in the global coordinate system: e : Where T is the coordinate transformation matrix, T T represents the transpose of the coordinate transformation matrix; to facilitate subsequent calculations, k e Blocked into Among them, i and j are the numbers of two nodes on unit e; The global coordinate system and the local coordinate system only rotate around the y-axis, so the coordinate transformation matrix T is: Where α is the distance from the x-axis to The angle of the axis; For the longitudinal beam grid, α=0°; for the transverse beam grid, α=270°; S34, by the element stiffness matrix k e Assemble the overall stiffness matrix; the overall stiffness matrix K is a 3n×3n order square matrix; The main sub-block K ii is obtained by superimposing the main sub-blocks of each relevant unit of node i, that is, The sub-block K im ,When i,m are related nodes, they are the corresponding sub-blocks of the unit connecting them, i.e. When i and m are non-related nodes, they are zero sub-blocks; S35, after being combined into the overall stiffness matrix, according to the original stiffness equation of the structure F = KΔ, Where F is the external force at each node, Δ is the displacement at each node, and the expressions are 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 Represents the external force column vector of node i, M xi ,F yi ,M zi are the external couple along the x direction, the external force along the y direction, and the external couple along 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 of node i along the x direction, the linear displacement along the y direction, and the angular displacement along the z direction; The overall stiffness matrix K is a 3n×3n square matrix. The constraint support condition is introduced. The hollow slab-beam bridge is composed of p slab beams. q groups of transverse beam grids are set. The structure is (p-1)q hinge points between beams, 2p nodes at the support, pq rigid nodes, and 2(p-1)q nodes on both sides of the hinge joint. For the 2p hinge nodes at the beam end supports, we have θ xi =0, v yi =0,M zi =0; for the hinges 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. 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 node where pq transverse and longitudinal members are connected, the expression is: xi =0, F yi =F Eyi , M zi =M Ezi ; In the above formula, F Eyi ,M Ezi The equivalent node load assigned to the node for the known bridge deck vehicle load obtained in step S1 is calculated as follows; The vehicle load on the bridge deck is distributed as equivalent node loads. Under the vehicle load, the longitudinal member is subjected to a concentrated force F in the y direction, and its fixed end force F F The calculation formula is as follows: in, is the fixed end shear force at end i, is the fixed end shear force at end j, is the fixed end bending moment at end i, is the bending moment at the fixed end of end j; F is the concentrated force in the y direction acting on this member, l is the length of this member, a and b are the distances from the concentrated force to i and j respectively; The equivalent node load on any node i is the sum of the fixed end forces of all the members connected to it as follows: Among them, M Exi 、F Eyi 、M Ezi are the equivalent nodal couple acting on node i along the x direction, the equivalent nodal shear force along the y direction, and the equivalent nodal couple along the z direction; 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 above 3n constraints are combined with the original stiffness equation of the structure to obtain the actual displacement Δ of each node; The rod end internal force F of member e is calculated by the following formula e , including the bending moment at the rod end in the x direction Shear force at the rod end in the y direction Bending moment at the rod end in z direction That is F e It is the sum of the rod end force generated by the fixed end force and the equivalent node load: where F Ei is the equivalent node load of node i, k e is the element stiffness matrix in the global coordinate system, Δ e is the unit displacement matrix in the global coordinate system, (k e Δ e ) i The rod end force at end i generated by the equivalent node load; the rod end bending moment in the x direction The bending moment distribution inside the member is obtained by combining the bending moment diagram; that is, 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 calculate the area [x i ,x i+1 ] of the moment envelope area; Among them A M (n,t,x i ) represents the area [x i ,x i+1 ]’s bending moment envelope area.

4. The method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement according to claim 1, characterized in that: In step S4, the curvature envelope area of ​​each stiffness identification area is calculated according to the rotation angle of the unit node using the following formula: A C (n,t,x i )=θ(n,t,x i+1 )-θ(n,t,x i ) where θ(n,t,x i ) is the x value of the nth hollow plate at time t i The corner at A C (n,t,x i ) is the area [x i ,x i+1 ] is the curvature envelope area.

5. The method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement according to claim 1 or 4, characterized in that: Step S5 adopts the following formula: in, is the uniform stiffness of the i-th test unit A(n,i) on the n-th hollow plate, and T is the total effective measurement time.

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

7. The method for identifying regional distribution stiffness of hollow slab bridges based on visual measurement according to claim 1, characterized in that: Step S7 uses an iterative method to update the beam grillage model parameters to obtain the stiffness of the damaged area. The specific steps are: S71, after obtaining the bending stiffness value under the assumption that the structure is intact and identifying the damage, the stiffness value of the intact area is averaged and assigned to all the longitudinal beam grids in the intact area; the expression is as follows: in, represents the average uniform stiffness of the intact area under the assumption that the structure is intact, is the intact region stiffness adopted for the first iteration; The bending stiffness value of the damaged area is equal to the stiffness value of this area under the intact assumption. in, represents the uniform stiffness of the damaged area (n,i) under the assumption that the structure is intact, is the stiffness of the damaged region (n,i) adopted in the first iteration; For torsional stiffness, it is considered to vary in proportion to bending stiffness; S72, using the stiffness value obtained in S71, update the unit stiffness matrix k e , repeat steps S34 to S36 to obtain the moment envelope area (A) obtained in the first iteration M (n,t,x i )) loop1 ; Using the curvature envelope area A C (n,t,x i ), for each region, find the uniform stiffness of the region (n,i) obtained in the first iteration S73, again averaging the stiffness values ​​of the intact regions and assigning the average stiffness values ​​to all intact regions. in, is the average uniform stiffness of the intact area obtained in the (i-1)th iteration, is the stiffness value of the intact area adopted in the i-th iteration; The bending stiffness value of the damaged area is equal to the stiffness value of this area calculated last time. in, is the uniform stiffness of the damaged area (n,i) obtained in the (i-1)th iteration, is the uniform stiffness of the damaged region (n,i) adopted in the i-th iteration. For torsional stiffness, it is considered to vary in proportion to bending stiffness; S74, using the stiffness value obtained in S73, update the unit stiffness matrix k e Repeat steps S34 to S36 to obtain the moment envelope area (A M (n,t,x i )) loopi ; Using the curvature envelope area A C (n,t,x i ), for each region, find the uniform stiffness of region (n,i) S75, repeat steps S73 to S74 until convergence. Where ∈ is the relative difference. In order to balance the calculation efficiency and accuracy, ∈0 is set to 0.0001; at this time, That is the final uniform stiffness calculation result of region (n,i).

Images