Simple arrangement method of cluster distributed optical fiber deflection monitoring for continuous girder bridge

Abstract

This invention discloses a simplified deployment method for distributed optical fiber deflection monitoring of continuous beam bridge clusters. The method involves deploying upper and lower edge strain-sensing optical cables on the side beams of the continuous beam bridge cluster, and only deploying a lower edge strain-sensing optical cable on the middle beam. The deflection of the side beams is obtained through the upper and lower edge strain monitoring data, and the upper edge strain data of the middle beam is reconstructed using the upper edge strain data of the side beams. Combined with the lower edge strain monitoring data of the middle beam, an accurate reconstruction of the middle beam deflection is achieved through a data-physics hybrid-driven continuous beam bridge deflection reconstruction method. This method not only solves the problems of large sensor usage, complex deployment, high construction costs, and poor economic efficiency of cluster monitoring caused by the need to deploy optical fibers simultaneously on the upper and lower edges of each beam in existing technologies, but also addresses the difficulty of deploying sensors on the upper edge of the middle beam due to bridge deck paving and the resulting significant construction damage. This effectively improves the robustness and data reconstruction capability of the monitoring system.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring of the operational safety of civil engineering structures, and relates to a method for monitoring the clustered distributed optical fiber deflection of continuous beam bridges. Background Technology

[0002] As a crucial component of modern transportation infrastructure, the structural health of continuous beam bridges during operation directly impacts traffic safety and service life. Deflection is a vital indicator for assessing the stress state and overall stiffness of beam bridges. Long-term, continuous, and accurate monitoring of deflection is a key means to prevent structural defects, assess remaining service life, and ensure operational safety.

[0003] Currently, deflection monitoring mainly employs a combination of traditional measurement methods and modern sensing technologies. Traditional methods, such as levels and total stations, offer high accuracy but require manual point-by-point measurements, making automated and real-time monitoring difficult. Furthermore, they are highly susceptible to environmental influences and unsuitable for long-term, large-scale monitoring of clustered bridges. In recent years, fiber optic sensing-based monitoring technologies have gradually become a research hotspot in structural health monitoring due to their advantages such as resistance to electromagnetic interference, high durability, and the ability to perform distributed measurements.

[0004] Distributed fiber optic sensing technology enables continuous, real-time strain monitoring along the fiber optic path by deploying sensing optical cables on or inside the structure. In existing technologies, distributed optical fibers are typically deployed simultaneously at the upper and lower edges of the main beam. The cross-sectional curvature is calculated using the strain difference between the upper and lower edges, and then the deflection is obtained by integration. However, in practical engineering, especially for monitoring continuous beam bridge clusters, this method has significant shortcomings: 1. Complex and costly sensor deployment: Optical fibers need to be laid at the upper and lower edges of each beam, resulting in a large amount of optical cable, long deployment period, and high construction difficulty, especially for the retrofit monitoring of existing bridges; 2. Difficulty in deploying sensors at the upper edge of the middle beam: In most bridge structures, the upper edge of the middle beam is often covered by the bridge deck pavement layer. Deploying sensors requires damaging the pavement layer, affecting structural integrity and complicating construction; 3. Data processing depends on complete strain information: Existing deflection identification methods usually require simultaneous acquisition of the strain at the upper and lower edges of the same cross section. If both are indispensable, the monitoring system has poor robustness. Once a local sensor fails, the deflection of that cross section cannot be calculated; 4. Poor scalability of cluster monitoring: For a cluster of multiple continuous beam bridges, if each bridge is deployed in the traditional way, the total cost will increase exponentially, resulting in poor economic efficiency and making it difficult to promote.

[0005] Therefore, there is an urgent need for a distributed fiber optic deflection monitoring method for continuous beam bridge clusters that can ensure monitoring accuracy while significantly reducing the number of sensors, simplifying the deployment process, and improving economic efficiency. Summary of the Invention

[0006] To address the challenges and high costs associated with deploying distributed fiber optic deflection monitoring sensors for continuous beam bridge clusters, this invention provides a simplified deployment method. This method involves deploying optical fibers only at the upper and lower edges of the side beams and the lower edge of the middle beam. It identifies the deflection using complete strain data from the side beams, reconstructs the strain at the upper edge of the middle beam by building a strain space relationship model, and finally achieves accurate identification of the middle beam deflection using a data-physics hybrid driving algorithm. This invention significantly reduces the amount of optical fiber used and the difficulty of deployment, lowers the cost of cluster monitoring, and ensures monitoring accuracy, making it suitable for long-term health monitoring of large-scale continuous beam bridge clusters.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] A simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges includes the following steps:

[0009] Step 1: According to the design and construction drawings of the monitoring system, deploy the upper and lower edge distributed strain sensing optical cables on the side beams of the continuous beam bridge in the cluster, and deploy only the lower edge distributed strain sensing optical cables on the middle beam of the continuous beam bridge. Collect the distributed fiber strain monitoring data of the upper and lower edge of the side beam of a certain continuous beam bridge A and the distributed fiber strain monitoring data of the lower edge of the middle beam of a certain continuous beam bridge in the cluster.

[0010] Step 2: Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in Step 1, the deflection data of beam A is calculated by span using the physical relationship model between distributed strain and deflection.

[0011] Step 3: Based on the distributed fiber optic strain monitoring data of the upper edge of beam A obtained in Step 1, the distributed fiber optic strain reconstruction data of the upper edge of beam B is obtained using the distributed strain and position data relationship model based on MS-CNN network.

[0012] Step 4: Based on the distributed fiber strain monitoring data of the lower edge of beam B obtained in Step 1 and the distributed fiber strain reconstruction data of the upper edge of beam B obtained in Step 3, the deflection reconstruction data of beam B is obtained using the continuous beam bridge deflection reconstruction method based on data-physics hybrid drive.

[0013] Step 5: For each continuous beam bridge in the cluster, repeat the process from Step 1 to Step 4 above to obtain the deflection monitoring data of the side beams and middle beams of all bridges in the cluster, thereby realizing distributed fiber optic deflection monitoring of the continuous beam bridge cluster.

[0014] Compared with the prior art, the present invention has the following advantages:

[0015] 1. This invention optimizes the deployment strategy of sensing optical cables by deploying distributed optical fibers only on the upper and lower edges of the side beams and the lower edge of the middle beam. It directly calculates the deflection of the side beams using complete strain data, and reconstructs the strain of the upper edge of the middle beam based on the strain data of the upper edge of the side beams and a neural network model. Finally, it combines a physical and mechanical model to achieve accurate and efficient identification of the deflection of the middle beam and the entire bridge.

[0016] 2. This invention significantly reduces the number of sensors used and the complexity of deployment, greatly reducing the economic cost and construction difficulty of cluster monitoring, while ensuring monitoring accuracy and system robustness. It provides an economical, reliable and scalable technical solution for structural health monitoring of large-scale continuous beam bridge clusters. Attached Figure Description

[0017] Figure 1 Finite element model diagram of a three-span continuous concrete T-beam bridge;

[0018] Figure 2 The deflection identification result for beam A;

[0019] Figure 3 To simulate the frequency histogram of vehicle weight;

[0020] Figure 4 The iterative convergence graph of the data relationship model between distributed strain and location;

[0021] Figure 5 The deflection reconstruction result of beam B;

[0022] Figure 6 The error between the reconstructed deflection value and the actual value of beam B;

[0023] Figure 7 A flowchart illustrating a simplified deployment method for distributed optical fiber deflection monitoring of continuous beam bridge clusters. Detailed Implementation

[0024] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but it is not limited thereto. Any modifications or equivalent substitutions to the technical solution of the present invention that do not depart from the spirit and scope of the technical solution of the present invention should be covered within the protection scope of the present invention.

[0025] This invention provides a simplified deployment method for distributed fiber optic deflection monitoring of continuous beam bridge clusters. The method involves deploying upper and lower edge strain-sensing optical cables on the side beams of the continuous beam bridge cluster, and only deploying a lower edge strain-sensing optical cable on the middle beam. The deflection of the side beams is obtained through the upper and lower edge strain monitoring data, and the upper edge strain data of the middle beam is reconstructed using the upper edge strain data of the side beams. Combined with the lower edge strain monitoring data of the middle beam, an accurate reconstruction of the middle beam deflection is achieved through a data-physical hybrid-driven continuous beam bridge deflection reconstruction method. Figure 7 As shown, the specific steps include the following:

[0026] Step 1: Based on the monitoring system design and construction drawings, deploy distributed strain sensing optical cables at the upper and lower edges of the side beams of the continuous beam bridges within the cluster, and only deploy distributed strain sensing optical cables at the lower edge of the middle beams of the continuous beam bridges. Collect distributed fiber optic strain monitoring data at the upper and lower edges of the side beam A and the lower edge of the middle beam B of a certain continuous beam bridge within the cluster.

[0027] Step Two: Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in Step One, the deflection data of beam A is calculated span by span using the physical relationship model between distributed strain and deflection. The specific steps are as follows:

[0028] Step Two One: Assume a certain The first continuous beam The span is The upper and lower edges of the A-span side beam are evenly distributed with The beam is divided along its length by several measuring points. Units, each unit length .in, Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in step one, at a certain moment, the set of distributed fiber optic strain monitoring data of the upper edge of beam A is collected as a vector. The distributed fiber optic strain monitoring data set at the lower edge of beam A is a vector. The definition is as follows:

[0029] (1)

[0030] (2)

[0031] In the formula, For the A side beam A dataset of distributed optical fiber strain monitoring across the upper edge. ; For the A side beam Data set of distributed optical fiber strain monitoring data at the lower edge of the straddle. and The definition is as follows:

[0032] (3)

[0033] (4)

[0034] In the formula, For the first Cross the first Distributed fiber optic strain monitoring data at the upper edge of beam A at each measuring point. ; For the first Cross the first Distributed fiber optic strain monitoring data at the lower edge of beam A at each measuring point.

[0035] Step 22: Obtaining the first edge beam of side A using distributed fiber optic strain sensors. Cross the first The average strain at the upper and lower edges of each unit and The distance between the upper and lower edge sensors is , No. Cross the first The average curvature of each unit is expressed as:

[0036] (5)

[0037] In the formula, For the A side beam Cross the first The average curvature of each unit .

[0038] Steps two and three: Using the physical relationship model of distributed strain and deflection, obtain the first... Span deflection data :

[0039] (6)

[0040] In the formula, For the A side beam Cross the first Deflection at each measuring point ; For bending stiffness; Shear modulus; This represents the effective shear area.

[0041] Step 3: Based on the distributed fiber optic strain monitoring data of the upper edge of beam A obtained in Step 1, the distributed fiber optic strain reconstructed data of the upper edge of beam B is obtained using a distributed strain-location data relationship model based on an MS-CNN network. The specific steps are as follows:

[0042] Step 31: Under a certain sampling, the set of distributed fiber strain simulation data of the upper edge of beam A obtained from the finite element model is a vector. The simulation data set of distributed optical fiber strain on the upper edge of beam B is a vector. The definition is as follows:

[0043] (7)

[0044] (8)

[0045] In the formula, For the A side beam A dataset of simulation data on strain in distributed optical fibers across the upper edge; For the middle beam of B A dataset of simulation data of strain in distributed optical fibers across the upper edge. and The definition is as follows:

[0046] (9)

[0047] (10)

[0048] In the formula, For the first Cross the first Simulation data of distributed fiber optic strain at the upper edge of beam A at each measuring point. , In the finite element model, the first Number of measuring points on the upper (lower) edge of the beam spanning A; For the first Cross the first Simulation data of distributed optical fiber strain at the upper edge of beam B at each measuring point.

[0049] Step 32: Employ a multi-scale convolutional neural network (MS-CNN) to extract spatial dimension features and construct a data relationship model between the upper edge distributed strain and position of the A side beam and B middle beam of the continuous beam bridge. For one-dimensional sequence data... Perform multiple convolution and pooling operations:

[0050] (11)

[0051] (12)

[0052] In the formula, For the first The characteristics of the strain simulation data of the upper edge of beam A are used as the input for convolution; convolution kernel The One parameter; convolution kernel The bias; convolution kernel Size; For the first One to the first The inputs are processed by the convolution kernel. Processed output characteristics; It is the ReLU activation function; Pooling layer No. Each output feature; Pooling layer No. The output feature corresponds to the window of the first... One value; This is the pooled window size.

[0053] After processing through multiple convolutional and pooling layers, the features extracted from the strain simulation data of the upper edge of beam A are as follows:

[0054] (13)

[0055] In the formula, For the last pooling layer The Each output feature; The feature vector of the upper edge strain simulation data of beam A extracted by the MS-CNN network.

[0056] The constructed MS-CNN network model is essentially a parametric regression model used to approximate the mapping function. As shown in the following formula:

[0057] (14)

[0058] (15)

[0059] In the formula, It is a multi-scale convolutional neural network model.

[0060] Step 33: Using the simulated strain data of the upper edge of beam B as the training target for the entire model output, the model is trained through the error backpropagation algorithm. The training objective of the MS-CNN network is to obtain the true value of the simulated strain data of the upper edge of beam B. With reconstructed value The mean squared error (MSE) is minimized as shown in the following equation:

[0061] (16)

[0062] In the formula, The loss function; For the middle beam of B Reconstructed vector of upper edge strain simulation data from subsampled samples ; For the middle beam of B The true vector of the upper edge strain simulation data from the second sampling; This represents the total number of samples.

[0063] Steps 3 and 4: Input the distributed fiber optic strain monitoring data of the upper edge of beam A of the continuous beam bridge collected in Step 1 into the trained distributed strain and position data relationship model to obtain the reconstructed distributed fiber optic strain data of the upper edge of beam B, as shown in the following formula:

[0064] (17)

[0065] In the formula, Let be the measured strain vector at the upper edge of beam A; Let be the upper edge strain reconstruction vector of beam B.

[0066] Step 4: Based on the distributed fiber optic strain monitoring data of the lower edge of beam B obtained in Step 1 and the distributed fiber optic strain reconstruction data of the upper edge of beam B obtained in Step 3, the deflection reconstruction data of beam B is obtained using the data-physics hybrid driven continuous beam bridge deflection reconstruction method. The specific steps are as follows:

[0067] Step 41: Obtain the distributed fiber optic strain monitoring data of the lower edge of beam B based on Step 1, and obtain the distributed fiber optic strain reconstruction data of the upper edge of beam B based on Step 3. Assume that at a certain moment, the set of distributed fiber optic strain monitoring data of the lower edge of beam B is a vector. The distributed fiber strain reconstruction data set at the upper edge of beam B is a vector. The definition is as follows:

[0068] (18)

[0069] (19)

[0070] In the formula, For the middle beam of B Data set of distributed optical fiber strain monitoring data at the lower edge of the straddle; For the middle beam of B A dataset of cross-upper edge distributed fiber strain reconstruction data. and The definition is as follows:

[0071] (20)

[0072] (twenty one)

[0073] In the formula, For the first Cross the first Distributed fiber optic strain monitoring data of the lower edge of beam B at each measuring point; For the first Cross the first Distributed fiber strain reconstruction data of the upper edge of beam B at each measuring point.

[0074] Step 42: Based on the reconstructed upper edge strain and the measured lower edge strain, the average curvature reconstruction vector of beam B can be obtained. :

[0075] (twenty two)

[0076] In the formula, Let be the measured vector of the average lower edge strain of beam B. ; Let B be the reconstructed vector of the average upper edge strain of beam B. .

[0077] Step 43: Using a data-physics hybrid-driven continuous beam bridge deflection reconstruction method, obtain the deflection of beam B. Span deflection reconstruction data :

[0078] (twenty three)

[0079] In the formula, For the B middle beam of the continuous beam bridge Cross the first Deflection at each measuring point .

[0080] Step 5: For each continuous beam bridge within the cluster, repeat steps 1 to 4 above to obtain deflection monitoring data for the side beams and middle beams of all bridges in the cluster, thus realizing distributed fiber optic deflection monitoring of the continuous beam bridge cluster. The specific steps are as follows:

[0081] Step 51: Collect distributed fiber optic strain monitoring data of the upper and lower edges of the A side beam and the lower edge of the B middle beam of other continuous beam bridges in the cluster;

[0082] Step 52: Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in Step 51, use the method in Step 2 to obtain the deflection data of beam A.

[0083] Step 53: Based on the distributed fiber optic strain monitoring data of the upper edge of beam A obtained in Step 51, use the method in Step 3 to obtain the distributed fiber optic strain reconstruction data of the upper edge of beam B.

[0084] Step 54: Based on the distributed fiber optic strain monitoring data of the lower edge of beam B obtained in Step 51 and the distributed fiber optic strain reconstruction data of the upper edge of beam B obtained in Step 53, the deflection reconstruction data of beam B is obtained using the method in Step 4.

[0085] The simplified deployment method for distributed optical fiber deflection monitoring of continuous beam bridge clusters described above not only solves the problems of large sensor usage, complex deployment, high construction costs, and poor economic efficiency of cluster monitoring caused by the need to simultaneously deploy optical fibers on the upper and lower edges of each beam in existing technologies, but also solves the problem of difficulty in deploying sensors on the upper edge of the middle beam due to bridge deck paving and significant construction damage. It effectively improves the robustness and data reconstruction capability of the monitoring system, and can still achieve accurate deflection identification even when some strain data is missing. At the same time, by constructing a data-physical hybrid driving model and integrating measured data with structural mechanical relationships, the accuracy and reliability of deflection identification are further improved.

[0086] The following experiments were conducted to verify the effectiveness of the present invention:

[0087] This experiment uses a 20+20+20m three-span continuous concrete T-beam bridge as an example. Two simulated test conditions were used: 1) With strain gauges simultaneously deployed at the upper and lower edges, to verify the effectiveness of the proposed distributed strain-deflection physical relationship model, a static loading condition was set on the #1 side span of the three-span continuous concrete T-beam bridge, using a concentrated force loading mode. Distributed optical fibers were deployed at the upper and lower edges of side beam A, and the distributed strain-deflection physical relationship model was used to identify the deflection of side beam A; 2) With strain gauges only deployed at the lower edge of the beam, to verify the effectiveness of the proposed data-physics hybrid-driven continuous beam bridge deflection reconstruction method, a static loading condition was set on the #1 side span of the three-span continuous concrete T-beam bridge, using a concentrated force loading mode. Distributed optical fibers were deployed at the lower edge of middle beam B, and the data-physics hybrid-driven continuous beam bridge deflection reconstruction method was used to reconstruct the deflection of middle beam B. Distributed optical fiber sensing technology was deployed with one strain gauge every 0.2m along the longitudinal direction of the bridge. The simulated test conditions are shown in Table 1, and the finite element model of the continuous beam bridge is as follows: Figure 1 As shown.

[0088]

[0089] The specific details of this experiment are as follows:

[0090] For working condition 1: Distributed optical fibers are deployed on the upper and lower edges of beam A. This invention uses a physical relationship model of distributed strain and deflection to obtain the deflection of beam A. First, the longitudinal bending curvature of span #1 is obtained using formula (1). Then, the deflection identification result of span #1 is obtained using formula (2) of the physical relationship model of distributed strain and deflection of continuous beams, as shown below. Figure 2 As shown in (a). Similarly, the deflection identification results for spans #2 and #3 are obtained sequentially, as shown in (a). Figure 2 As shown in (b) and 2(c). Finally, the results of the span identification are integrated into the overall deflection identification results of the continuous beam bridge, as shown in... Figure 2As shown in (d). To compare the effectiveness of the algorithms, the classical conjugate beam method (Method 1) and the conjugate beam method considering shear deformation (Method 2) were compared.

[0091] For scenario two, where distributed optical fibers are only deployed at the lower edge of beam B, this paper employs a data-physical hybrid-driven continuous beam bridge deflection reconstruction method to obtain the deflection of beam B. First, using the spatial relationship between the strain of side beam A and the upper edge of beam B, a data relationship model of distributed strain and location is constructed to reconstruct the strain distribution at the upper edge of beam B. Then, the measured lower edge strain of beam B and the reconstructed upper edge strain are input together into the physical relationship model of distributed strain and deflection to calculate the deflection of beam B.

[0092] To construct a data relationship model between the distributed strain and spatial position of the upper edges of the A-side beam and the B-middle beam, random vehicle loading conditions during operation were simulated on the #1 side span of a three-span continuous concrete T-beam bridge. The simulation results of 100 static loading cycles using a two-axle vehicle were used to build the relational database. The vehicle weight ranged from 3.5 to 28.62 tons. The frequency distribution of the 100 random vehicle weight cycles was obtained using the Monte Carlo method, as shown below. Figure 3 As shown. Using 100 generated random load samples, a three-span continuous concrete T-beam bridge was statically loaded sequentially. Distributed strain sensor data were collected at the upper edges of the A-side beam and B-middle beam during each loading simulation. The first 80 sets of strain sensor data from the upper edges of the A-side beam and B-middle beam were divided into a training set and the last 20 sets into a validation set. The strain sensor data from the upper edges of the A-side beam were used as input, and the strain sensor data from the upper edges of the B-middle beam were used as output to train the model. The model iteratively converged as shown. Figure 4 As shown.

[0093] The strain sensor data at the upper edge of beam A, collected under working condition 2, is input into the trained distributed strain-spatial position data relationship model to obtain the output strain sensor reconstruction data at the upper edge of beam B. The reconstructed strain data at the upper edge of beam B and the measured strain data at the lower edge of beam B are then input into the distributed strain-deflection physical relationship model. The reconstructed deflection of beam B is calculated using a data-physics hybrid-driven continuous beam bridge deflection reconstruction method. Figure 5 As shown. The actual deflection is obtained by inputting the measured strain values ​​of the upper and lower edges of beam B into the physical relationship model of distributed strain and deflection. The deflection error of beam B reconstructed based on data-physics hybrid driving is as follows: Figure 6 As shown.

Claims

1. A simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges, characterized in that... The method includes the following steps: Step 1: According to the design and construction drawings of the monitoring system, deploy the upper and lower edge distributed strain sensing optical cables on the side beams of the continuous beam bridge in the cluster, and deploy only the lower edge distributed strain sensing optical cables on the middle beam of the continuous beam bridge. Collect the distributed fiber strain monitoring data of the upper and lower edge of the side beam of a certain continuous beam bridge A and the distributed fiber strain monitoring data of the lower edge of the middle beam of a certain continuous beam bridge in the cluster. Step Two: Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in Step One, the deflection data of beam A is calculated span by span using the physical relationship model between distributed strain and deflection. The specific steps are as follows: Step Two One: Assume a certain The first continuous beam The span is The upper and lower edges of the beam spanning A are evenly distributed with The beam is divided along its length by several measuring points. Units, each unit length ,in, Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in step one, at a certain moment, the set of distributed fiber optic strain monitoring data of the upper edge of beam A is collected as a vector. The distributed fiber optic strain monitoring data set at the lower edge of beam A is a vector. The definition is as follows: In the formula, For the A side beam A dataset of distributed optical fiber strain monitoring across the upper edge. ; For the A side beam Data set of distributed optical fiber strain monitoring data at the lower edge of the straddle; Step 22: Obtaining the first edge beam of side A using distributed fiber optic strain sensors. Cross the first The average strain at the upper and lower edges of each unit and The distance between the upper and lower edge sensors is , No. Cross the first The average curvature of each unit is expressed as: In the formula, For the A side beam Cross the first The average curvature of each unit ; Steps two and three: Using the physical relationship model of distributed strain and deflection, obtain the first... Span deflection data : In the formula, For the A side beam Cross the first Deflection at each measuring point ; For bending stiffness; Shear modulus; Effective shear area; Step 3: Based on the distributed fiber optic strain monitoring data of the upper edge of beam A obtained in Step 1, the distributed fiber optic strain reconstruction data of the upper edge of beam B is obtained using the distributed strain and position data relationship model based on MS-CNN network. Step 4: Based on the distributed fiber strain monitoring data of the lower edge of beam B obtained in Step 1 and the distributed fiber strain reconstruction data of the upper edge of beam B obtained in Step 3, the deflection reconstruction data of beam B is obtained using the continuous beam bridge deflection reconstruction method based on data-physics hybrid drive. Step 5: For each continuous beam bridge in the cluster, repeat the process from Step 1 to Step 4 above to obtain the deflection monitoring data of the side beams and middle beams of all bridges in the cluster, thereby realizing distributed fiber optic deflection monitoring of the continuous beam bridge cluster.

2. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 1, characterized in that... In step two, and The definition is as follows: In the formula, For the first Cross the first Distributed fiber optic strain monitoring data at the upper edge of beam A at each measuring point. ; For the first Cross the first Distributed fiber optic strain monitoring data at the lower edge of beam A at each measuring point.

3. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 1, characterized in that... The specific steps of step three are as follows: Step 31: Under a certain sampling, the set of distributed fiber strain simulation data of the upper edge of beam A obtained from the finite element model is a vector. The simulation data set of distributed optical fiber strain on the upper edge of beam B is a vector. The definition is as follows: In the formula, For the A side beam A dataset of simulation data on strain in distributed optical fibers across the upper edge; For the middle beam of B A dataset of simulation data on strain in distributed optical fibers across the upper edge; Step 32: Employ a multi-scale convolutional neural network (MS-CNN) to extract spatial dimension features, constructing a data relationship model of the upper edge distributed strain and position of the A side beam and B middle beam of the continuous beam bridge. This is for one-dimensional sequence data. Perform multiple convolution and pooling operations: In the formula, For the first The characteristics of the strain simulation data of the upper edge of beam A are used as the input for convolution; convolution kernel The One parameter; convolution kernel The bias; convolution kernel Size; For the first One to the first The inputs are processed by the convolution kernel. Processed output characteristics; It is the ReLU activation function; Pooling layer No. Each output feature; Pooling layer No. The output feature corresponds to the window of the first... One value; This is the pooling window size; After processing through multiple convolutional and pooling layers, the features extracted from the strain simulation data of the upper edge of beam A are as follows: In the formula, For the last pooling layer The Each output feature; The feature vector of the upper edge strain simulation data of beam A extracted by the MS-CNN network; Step 33: Using the strain simulation data of the upper edge of beam B as the training target for the entire model output, the model is trained through the error backpropagation algorithm; Steps 3 and 4: Input the distributed fiber optic strain monitoring data of the upper edge of beam A of the continuous beam bridge collected in Step 1 into the trained distributed strain and position data relationship model to obtain the reconstructed distributed fiber optic strain data of the upper edge of beam B, as shown in the following formula: In the formula, Let be the measured strain vector at the upper edge of beam A; Let be the upper edge strain reconstruction vector of beam B.

4. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 3, characterized in that... In step three, and The definition is as follows: In the formula, For the first Cross the first Simulation data of distributed fiber optic strain at the upper edge of beam A at each measuring point. , In the finite element model, the first Number of measuring points at the upper / lower edges of the beam spanning A; For the first Cross the first Simulation data of distributed optical fiber strain at the upper edge of beam B at each measuring point.

5. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 3, characterized in that... In step 3.2, the constructed MS-CNN network model is essentially a parametric regression model used to approximate the mapping function. As shown in the following formula: In the formula, It is a multi-scale convolutional neural network model.

6. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 3, characterized in that... In step three, the training objective of the MS-CNN network is to obtain the true values ​​of the simulated strain data at the upper edge of beam B. With reconstructed value Minimize the mean squared error, as shown in the following equation: In the formula, The loss function; For the middle beam of B Reconstructed vector of upper edge strain simulation data from subsampled samples ; For the middle beam of B The true vector of simulated upper edge strain data from the second sampling; This represents the total number of samples.

7. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 3, characterized in that... The specific steps of step four are as follows: Step 41: Assume that at a certain moment, the distributed fiber optic strain monitoring data set at the lower edge of beam B is a vector. The distributed fiber strain reconstruction data set at the upper edge of beam B is a vector. The definition is as follows: In the formula, For the middle beam of B Data set of distributed optical fiber strain monitoring data at the lower edge of the straddle; For the middle beam of B Cross-upper edge distributed fiber strain reconstruction dataset; Step 42: Obtain the average curvature reconstruction vector of beam B based on the reconstructed upper edge strain and the measured lower edge strain. : In the formula, Let be the measured vector of the average lower edge strain of beam B. The average upper edge strain reconstruction vector of beam B; Step 43: Obtain the middle beam of B according to the following formula. Span deflection reconstruction data : In the formula, For the B middle beam of the continuous beam bridge Cross the first Deflection at each measuring point .

8. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 7, characterized in that... In step four one, and The definition is as follows: In the formula, For the first Cross the first Distributed fiber optic strain monitoring data of the lower edge of beam B at each measuring point; For the first Cross the first Distributed fiber strain reconstruction data of the upper edge of beam B at each measuring point.

9. The simplified deployment method for clustered distributed optical fiber deflection monitoring of continuous beam bridges according to claim 7, characterized in that... The specific steps of step five are as follows: Step 51: Collect distributed fiber optic strain monitoring data of the upper and lower edges of the A side beam and the lower edge of the B middle beam of other continuous beam bridges in the cluster; Step 52: Based on the distributed fiber optic strain monitoring data of the upper and lower edges of beam A obtained in Step 51, use the method in Step 2 to obtain the deflection data of beam A. Step 53: Based on the distributed fiber optic strain monitoring data of the upper edge of beam A obtained in Step 51, use the method in Step 3 to obtain the distributed fiber optic strain reconstruction data of the upper edge of beam B. Step 54: Based on the distributed fiber optic strain monitoring data of the lower edge of beam B obtained in Step 51 and the distributed fiber optic strain reconstruction data of the upper edge of beam B obtained in Step 53, the deflection reconstruction data of beam B is obtained using the method in Step 4.

Images