Bridge dynamic deflection and dynamic strain data fusion method
By deploying deflection and strain measuring points at the bottom of the bridge, and using matrix methods and interpolation differentiation to establish a mechanical coupling model of dynamic deflection and dynamic strain, the problems of time-domain reference deviation and nonlinear vibration error in bridge dynamic deflection and dynamic strain monitoring are solved, and the accurate fusion of multi-physics data and real-time reliable evaluation are realized.
Patent Information
- Application Number
- CN202511311105.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-15
- Publication Date
- 2025-11-21
- Estimated Expiration
- 2045-09-15
AI Technical Summary
In existing bridge dynamic deflection and strain monitoring technologies, the time-domain reference deviation is caused by the difference in sampling frequency of multi-source heterogeneous data. Traditional fusion methods experience a surge in error in nonlinear vibration, and single-parameter analysis fails to reveal the mechanical coupling mechanism between dynamic deflection and dynamic strain, resulting in missed detection of structural damage characteristics and inaccurate assessment.
A matrix method was used to deploy deflection and strain measuring points at the bottom of the bridge. A dynamic deflection-dynamic strain mechanical coupling model was established by cubic spline interpolation and quadratic differentiation. Data fusion units were constructed and spatial corrections were performed. Torsional deformation was ignored, and a probabilistic statistical model based on measured data was established for data fusion.
It achieves precise spatiotemporal matching of multi-physics field data, reduces the missed detection rate of structural damage features, supports early warning of micro-damage, reduces the overall cost of monitoring systems, is suitable for synchronization of heterogeneous data at low, medium and high frequencies, and improves the real-time reliability of bridge dynamic response.
Smart Images

Figure CN120800711B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the field of bridge health monitoring, and relates to a bridge dynamic deflection and dynamic strain data fusion method. BACKGROUND
[0002] In the field of bridge health monitoring, dynamic deflection and dynamic strain are core parameters for evaluating structural dynamic response. The existing technology generally adopts a separate collection and analysis mode: dynamic deflection data is obtained by a laser displacement meter or acceleration integration method, and dynamic strain data is independently collected by a resistance strain gauge and the like. However, due to the difference in sampling frequency (up to 10-100 times) between the two types of data, the time reference deviates. The traditional interpolation correction method produces a phase error of 0.5%-2% under non-stationary vibration, which weakens the dynamic response correlation. At the same time, the existing analysis adopts a single parameter threshold warning, which fails to reveal the mechanical coupling mechanism between dynamic deflection and dynamic strain, resulting in 15%-30% of structural damage feature missing; and the traditional fusion method (such as Kalman filtering) is limited by the linear assumption, and the error increases by more than 40% in nonlinear vibration with an amplitude greater than 5 mm.
[0003] In the current solution, a high-precision synchronous collection system can alleviate the data asynchrony, but the hardware cost increases by 3-5 times and the sampling rate is limited (≤200 Hz), and the physical model driven method fails more than 60% of the model due to bridge material degradation. The above defects show that the existing technology cannot realize the deep correlation of multi-physical field data and the accurate fusion of nonlinear dynamic characteristics, which restricts the reliability and real-time performance of bridge safety evaluation.
[0004] In view of the core defects in the existing bridge dynamic deflection and dynamic strain monitoring technology, the technical problems to be solved include: 1) how to eliminate the time domain reference deviation of multi-source heterogeneous data caused by the difference in sampling frequency (10-100 times) without relying on high-cost synchronous hardware; 2) how to break through the limitation of single parameter analysis, establish a nonlinear mechanical coupling relationship model between dynamic deflection and dynamic strain, and enhance the dynamic response correlation; 3) how to build a new fusion framework that fuses structural mechanics constitutive and data-driven, overcome the defect that the error of the traditional linear fusion algorithm (such as Kalman filtering) increases sharply in nonlinear vibration, and avoid the model failure of the physical model driven method due to material degradation.
[0005] The application adopts a matrix method to reveal the mechanical coupling mechanism between dynamic deflection and dynamic strain, and accurately establishes a corresponding calculation model. SUMMARY
[0006] The application aims to provide a bridge dynamic deflection and dynamic strain data fusion method to solve the problems listed in the background and realize the spatiotemporal accurate matching of multi-physical field data, the deep correlation of nonlinear dynamic characteristics and the real-time reliable evaluation of bridge dynamic response.
[0007] Technical scheme: A bridge dynamic deflection and dynamic strain data fusion method, deflection measuring points and strain measuring points are arranged on the rectangular grid formed in the longitudinal and transverse directions at the bottom of the bridge; the span direction of the bridge is taken as the longitudinal direction, and the bridge cross section perpendicular to the span direction of the bridge is taken as the position reference of the data fusion point, thereby determining the deflection data and strain data of different position points, and the selection and calculation of the data fusion point include:
[0008] (1) If there are strain measuring points and deflection measuring points on the bridge cross section, the strain measuring points on the bridge cross section are taken as the data fusion point, and the deflection data contained in the longitudinal section of the bridge is directly fused;
[0009] (2) If there are only strain measuring points on the bridge cross section, the strain measuring points on the bridge cross section are selected as the data fusion point, the deflection data is calculated by cubic spline interpolation with the boundary condition that the longitudinal strain of the bridge at the support is approximately 0, and then the fusion of the deflection data and the strain data is performed;
[0010] (3) If there are only deflection measuring points on the bridge cross section, the deflection measuring point position on the section is selected as the data fusion point, the curve expression is constructed by cubic spline interpolation of the measured data of several deflection measuring points on the same beam bottom, and the strain data of the data fusion point is obtained based on the quadratic derivation of the curve expression, and then the fusion of the deflection data and the strain data is performed.
[0011] In the method, for more than one deflection measuring point on the bridge cross section, when the deflection of the data fusion point on the non-deflection monitoring longitudinal section is obtained, a spatial correction term is added to the interpolation deflection expression to calculate the deflection data of the data fusion point;
[0012] The spatial correction term is determined by cubic spline interpolation method according to the deflection data of the deflection measuring points included in the matrix network, if the four measuring points corresponding to the matrix network are represented as A( , , ), B( , , ), C( , , ), D( , , ), then the spatial correction term is calculated by the following formula :
[0013] .
[0014] Further, in the method, for more than one strain measuring point existing on the bridge cross section, strain data of one of the strain measuring points is selected for data fusion.
[0015] The data fusion mode is to construct a fusion unit based on a rectangular network and a determined data fusion point, with adjacent data fusion points in the longitudinal bridge direction as two ends of the data fusion unit; then the fusion unit is expressed as a unit geometric matrix, the deflection corresponds to a bridge displacement column vector, and the strain corresponds to a bridge deformation column vector, and the positioning vectors are array integrated to form an overall geometric pseudo-matrix and a deformation pseudo-matrix for data fusion and storage.
[0016] Among them, for high-frequency data with a sampling frequency greater than 1Hz, the two ends of the data fusion unit are adjacent data fusion points in the longitudinal bridge direction, and a coordinate system is established on the two-dimensional plane at the top of the bridge, the deflection difference at the two ends of the constructed data fusion unit and the unit length are used to calculate the rotation angle, and the rotation angle and the unit length are used to calculate the strain, and the calculation formula is expressed as:
[0017]
[0018] In the formula, is the rotation angle, denotes the longitudinal bridge direction displacement of the end of the data fusion unit, denotes the longitudinal bridge direction displacement of the head of the data fusion unit, and is the representation of deflection data, denotes the elongation deformation of the bridge along the axis, which is the change representation of the strain force, and the calculation of the data fusion point is:
[0019] .
[0020] Further, the method ignores the transverse bridge direction displacement, does not establish a data fusion unit between adjacent data fusion points in the transverse bridge direction, only considers the longitudinal bridge direction dynamic strain data, ignores the torsional deformation, and considers that the change of the deflection data in the beam bottom area is continuous.
[0021] Further, the method requires that there be strain measuring points at both ends of the bridge longitudinal section where the data fusion point is located, and that strain sensors be deployed.
[0022] Beneficial effects: compared with the prior art, the method provided by the application does not rely on a complex machine learning model, but performs partition interval probability statistical modeling based on measured data characteristics, directly reconstructs a continuous time sequence through dynamic sampling, has advantages such as simplified simulation steps, high calculation efficiency, and clear physical meaning, and the specific advantages are as follows:
[0023] 1. This invention optimizes the deployment of deflection and strain measurement points, and then reconstructs the time reference of dynamic deflection-dynamic strain monitoring data based on time inversion algorithm considerations. It is suitable for low, medium and high frequency heterogeneous data synchronization, overcomes the strict requirements of traditional methods on data frequency matching, and is a universal algorithm architecture.
[0024] 2. This invention can effectively correlate dynamic deflection abrupt change with dynamic strain hysteresis effect, significantly reducing the missed detection rate of structural damage characteristics, while supporting early warning of micro-damage.
[0025] 3. By adjusting the number of data points, the calculation efficiency and accuracy can be flexibly balanced, and human experience parameters can be intervened to achieve controllable accuracy adjustment.
[0026] 4. The software-layer data fusion approach replaces the hardware-layer synchronization approach, which reduces the overall cost of the monitoring system and supports real-time processing of distributed sensor networks, making it applicable to a wide range of scenarios. Attached Figure Description
[0027] Figure 1 This is a flowchart of one embodiment of the present invention;
[0028] Figure 2 It is an array diagram of deflection and strain measurement points at the bottom of the bridge;
[0029] Figure 3 It is a displacement deformation element diagram;
[0030] Figure 4 This is a set of measured deflection data obtained in the embodiments;
[0031] Figure 5 This is a set of measured strain data obtained in the embodiments;
[0032] Figure 6 These are the original strain-time monitoring data from the embodiments. Detailed Implementation
[0033] The implementation and application of the present invention will be further described below with reference to the accompanying drawings.
[0034] First, the implementation process of the bridge dynamic deflection and dynamic strain data fusion method provided by the present invention includes the following steps:
[0035] S1. Optimize the deployment structure of strain sensors and deflection sensors on the bridge.
[0036] First, the longitudinal direction of the bridge is defined as its span, and the direction perpendicular to the longitudinal direction is defined as the transverse direction. Then, strain and deflection measuring points are deployed at the bottom of the bridge according to a network matrix. The sensors are all located at the bottom of the beam, and the beam height is only calculated. Considering the same, ignoring the longitudinal displacement of the bridge, only taking the longitudinal dynamic strain data, ignoring the torsional deformation, and considering that the deflection data in the beam bottom area is continuous.
[0037] In order to increase practicability, considering that a single piece of beam often has only one deflection sensor in the same cross section, and the transverse connection condition between multiple pieces of beams in the actual application scene is difficult to determine, the embodiment does not establish a unit between the adjacent data fusion points in the transverse direction of the bridge.
[0038] S2, determination of data of different types of fusion points.
[0039] The bridge dynamic deflection and dynamic strain fusion to be solved by the application is established in a coordinate system with the bridge bottom as a two-dimensional plane, corresponding to the longitudinal direction of the bridge and the transverse direction of the bridge. Then, the bridge cross section is taken as a reference for the position of the data fusion point, and the deployment of the sensor is considered to be at the bridge bottom position, so that there may be a strain measuring point or a deflection measuring point at the tangent of the bridge cross section and the bridge bottom. The specific conditions include:
[0040] 1) In the longitudinal position, the principle of minimizing interpolation is adopted, if the longitudinal strain and deflection measuring point are in the same section, the strain monitoring position of the section is selected, and interpolation is not needed.
[0041] 2) The strain monitoring position should be selected, and the deflection is obtained by cubic spline interpolation with the natural boundary condition that the longitudinal strain of the beam at the support is approximately 0.
[0042] 3) For the selected deflection monitoring position, the strain is obtained by the cubic spline interpolation deflection curve expression , and then the second derivative of the expression is obtained , that is , is a function of the height of the beam.
[0043] 4) For the positions without deflection data and strain data, generally, the data at the position has no practical significance and will also cause a large amount of calculation, and should be discarded. However, for the application, the calculation and derivation can be performed in combination with the above 2) and 3), and it is required that strain sensors should be distributed on both sides of the bridge cross section.
[0044] After fusion, each data fusion point has deflection and strain data.
[0045] S3, strain data and deflection data fusion based on the bridge longitudinal section.
[0046] In combination with the description of the application, the application is not limited to the details of the application described above, and various modifications and changes can be made without departing from the spirit and scope of the application. Figure 2 Figure 2 The strain measuring point and deflection measuring point arrangement structure of the bottom of the two bridges are included, on the bottom of the upper bridge, the bridge longitudinal strain measuring point and deflection measuring point are located in the same bridge longitudinal section, then are directly fused, on the bottom of the lower bridge, only the deflection measuring point is arranged on the bridge longitudinal section, and the data fusion point needs to be inferred from the deflection data.
[0047] In step S3, if there are multiple deflection measuring points in the transverse direction of the same longitudinal section, when obtaining the deflection of the data fusion point on the non-deflection monitoring longitudinal section, a spatial correction term needs to be added to the interpolation deflection expression The spatial correction term determines the dynamic deflection and dynamic strain data of the data fusion point by using the interpolation method of the surrounding sensors:
[0048] Suppose that the four corner points of the rectangular grid where the target point is located are measured (A, B, C, D)
[0049] The four measuring points are represented as A( , , ), B( , , ), C( , , ), D( , , ), then the spatial correction term is calculated by the following formula :
[0050] .
[0051] In the actual scene, the end of the box girder of the bridge is generally not provided with measuring points, but considering the replacement of new and old sensors, there can be multiple sensors on the bridge cross section, and thus for the present application, the following considerations exist:
[0052] In other embodiments, for high-frequency dynamic deflection data, frequent interpolation operation has too large computational load, therefore for the case of dense deflection measuring points, the following method can be used to improve the interpolation operation time interval according to the measured data, and the strain is solved by using the matrix method within the time interval, specifically:
[0053] Each displacement degree of freedom can establish an equation, the total number of node degrees of freedom is N, each unit has three unknown deformation parameters, and the total number of unknown deformation parameters M of the whole structure is 3 times the number of units.
[0054] In the global coordinate system, the deflection data in the data fusion unit is represented by the column displacement vector , e represents the data fusion unit number, and the longitudinal bridge displacement is approximately considered as 0, Vertical displacement of the bridge, For rotation, only in the vertical plane containing the unit axis, which can be integrated from the strain curve, there is the following expression:
[0055]
[0056] and respectively represent the column displacement vectors at the two ends of the data fusion unit, corresponding to the longitudinal displacement of the two ends of the bridge, and respectively correspond to the longitudinal displacement of the first and last ends of the data fusion unit. and respectively correspond to the first and last ends of the data fusion unit.
[0057] The corresponding deformation vector in the data fusion unit is represented as , , is the elongation deformation along the axis, which can be derived from the deflection interpolation curve, is the rotation of the first end of the data fusion unit relative to the new axis, is the rotation of the last end relative to the new axis. Let be 0 or derived from the deflection interpolation curve, and the total rotation is obtained by solving the equation set. The strain of the data fusion point is calculated as shown in Figure 3 .
[0058] Each longitudinal and transverse bridge adjacent data fusion point is considered as the two ends of the unit, then the vertical displacement and axial isometric deformation of the two ends of the unit are known, and the remaining elements are calculated according to the matrix, is the displacement vector of the unit rod end, is the unit geometry matrix, ; the unit geometry matrix equation in the global coordinate system is .
[0059] Deflection integration: the calculated single element displacement column vector is integrated in array (the k end matrix of the unit displacement vector is collected, (k=1,2, e is the unit number), and the integration rule is as follows: , (k, e) is the unit end positioning vector, k is the local code, and e is the global code).
[0060] Deflection integration: the calculated single element deformation column vector is integrated in array, and the integration rule is as follows: , (e) is the positioning vector, e is the global code, The whole deformation pseudo-matrix has the same column number as the cell number, so that the full-bridge strain is stored in the memory.
[0061] Combining Figure 2 and Figure 4 , the longitudinal position is determined, and the principle of minimizing interpolation is followed. If the longitudinal strain and deflection measuring points are in the same section, the strain monitoring position of the section is selected without interpolation, such as the upper half of e=1, 3, 5 and k=2 section. Then the data of the measured points in Figure 4 is stored in the space corresponding to the data fusion point, as shown in Figure 5 .
[0062] For the case of lacking measured data, Table 1 shows the deflection measurement at a certain time. The deflection is obtained by cubic spline interpolation with the natural boundary condition that the longitudinal strain of the beam at the support is approximately 0, as shown in Figure 6 , such as the upper half of e=1, 3, 5 and k=1 and e=6 and k=2 section.
[0063] Table 1. Deflection measurement
[0064] Beam span position (cm) 0 4625 9250 13875 18500 Deflection (cm) 0 -4.14 -11.67 -5.01 0
[0065] As shown in Table 2, the strain is obtained by twice differentiating the deflection expression of cubic spline interpolation, such as the lower half of e=1, 3, 5 and k=2 section.
[0066] Table 2. Strain obtained by twice differentiation
[0067] Beam span position (cm) 0 4625 9250 13875 18500 Deflection (με) 0 42 51 46 0
[0068] For the lower half of e=1, 3, 5 and k=1 and e=6 and k=2 section, both the deflection and strain measured data are lacking, and the data of Table 1 and Table 2 are obtained by the deflection expression of cubic spline interpolation and its twice differentiation.
[0069] For the lower half of e=2, the measured data at a certain time is shown in Table 3.
[0070] Table 3. Strain obtained by twice differentiation
[0071] Length l (cm) k = 1 deflection v1 (cm) k = 2 deflection v2 (cm) 2312 -3.6 -6.8
[0072] , .
Claims
1. A method for fusing dynamic deflection and dynamic strain data of bridges, characterized in that, Deflection and strain measurement points are deployed at the bottom of the bridge in a rectangular grid pattern along the longitudinal and transverse directions. With the bridge's span direction as the longitudinal direction, the cross-section perpendicular to the span direction is used as the reference for data fusion point locations. This determines the deflection and strain data at different locations. The selection and calculation of the data fusion points include: (1) If there are strain measurement points and deflection measurement points on the cross section of the bridge, the strain measurement points on the cross section of the bridge are used as data fusion points to directly fuse the deflection data contained on the cross section of the bridge. (2) If there are only strain measurement points on the cross section of the bridge, select the strain measurement points on the cross section of the bridge as data fusion points. The deflection data is obtained by cubic spline interpolation with the boundary condition that the longitudinal strain of the bridge at the support is approximately 0. Then, the deflection data and strain data are fused. (3) If there are only deflection measurement points on the cross section of the bridge, select the location of the deflection measurement point on the cross section as the data fusion point. Construct a curve expression by cubic spline interpolation of the measured data of several deflection measurement points on the bottom of the same beam, and perform a second derivative based on the curve expression to obtain the strain data of the data fusion point. Then perform the fusion of deflection data and strain data. In this method, when there is more than one deflection measurement point on the cross section of the bridge, when obtaining the deflection corresponding to the data fusion point on the non-deflection monitoring cross section, a spatial correction term is added to the interpolated deflection expression to calculate the deflection data of the data fusion point. The spatial correction term is determined using cubic spline interpolation based on the deflection data of the deflection measurement points included in the matrix network. If the four measurement points corresponding to the matrix network are represented as A( , , ), B( , , ), C( , , ),D( , , Then the space correction term is calculated using the following formula. : ; If there are more than one strain measurement point on the cross section of the bridge, the strain data of one strain measurement point is selected for data fusion. The data fusion method described above is based on constructing fusion units using a rectangular network and defined data fusion points, with adjacent data fusion points along the longitudinal direction of the bridge as the two ends of the data fusion unit. Then, the fusion unit is represented as a unit geometric matrix, where deflection corresponds to the bridge displacement column vector and strain corresponds to the bridge deformation column vector. The data is then arrayed and integrated according to the positioning vectors to form an overall geometric pseudo-matrix and a deformation pseudo-matrix for data fusion and storage.
2. The method for fusing bridge dynamic deflection and dynamic strain data according to claim 1, characterized in that, For high-frequency data with a sampling frequency greater than 1Hz, adjacent data fusion points along the longitudinal direction of the bridge are taken as the two ends of the data fusion unit, and a coordinate system is established with the top of the bridge as the two-dimensional plane. The rotation angle is calculated using the deflection difference between the two ends of the constructed data fusion unit and the unit length, and the strain is calculated using the rotation angle and the unit length. The calculation formula is expressed as follows: , In the formula, For the corner, This indicates the longitudinal displacement of the end bridge of the data fusion unit. This indicates the longitudinal displacement of the first end of the data fusion unit. and It is a representation of deflection data. This represents the elongation deformation of the bridge along its axis, which is a characterization of the change in strain force, and thus the data fusion points are obtained. The calculation is as follows: 。 3. The method for fusing bridge dynamic deflection and dynamic strain data according to claim 1, characterized in that, The method ignores transverse bridge displacement, does not establish data fusion units between adjacent data fusion points in the transverse bridge direction, only considers longitudinal bridge dynamic strain data, ignores torsional deformation, and assumes that the change of deflection data in the beam bottom region is continuous.
4. The method for fusing bridge dynamic deflection and dynamic strain data according to claim 1, characterized in that, This method requires strain measurement points to exist before and after the longitudinal position of the bridge cross section where the data fusion point is located, and strain sensors to be deployed.
Citation Information
Patent Citations
Large-span bridge girder dynamic displacement measurement system and data analysis method thereof
CN114812397A
Railway bridge structure health monitoring method based on data fusion
CN115452282A