Multi-channel distributed optical fiber sensing system

Abstract

The invention relates to the technical field of optical fiber sensing, and discloses a multichannel distributed optical fiber sensing system, comprising a bridge data collection module used for collecting bridge data; the two-dimensional structure data generation module is used for generating two-dimensional structure data from the collected bridge data, and the two-dimensional structure data comprises a data matrix and a relation matrix; the health state judgment module inputs the two-dimensional structure data into a judgment model, the judgment model outputs a health vector, and one component of the health vector represents the health state of one bridge key part; according to the method, the vibration, temperature and voiceprint data of the key parts of the bridge can be acquired and analyzed by collecting the bridge data regularly, the connection state between the key parts of the bridge and vibration noise interference introduced by traffic loads are fully considered, and compared with judgment of the health state of the bridge through a single part, the judgment accuracy is improved. The system can consider the historical state and dependency relationship of each bridge key part at the same time, and comprehensively reflect the health state of the whole bridge.

Description

Technical Field

[0001] The present invention relates to the field of optical fiber sensing technology, and more specifically, to a multi-channel distributed optical fiber sensing system. Background Art

[0002] As a key transportation infrastructure, the health of bridges is related to traffic safety and maintenance economy. Therefore, it is necessary to ensure their healthy operation through regular monitoring and timely detection of problems. Deep learning technology is used to analyze the vibration, temperature and soundprint information of key parts of the bridge obtained by the distributed fiber optic sensing system to evaluate the health status of the bridge. However, when vehicles pass through the bridge deck, the bridge will produce a vibration response due to the vehicle's mass, driving speed and uneven road surface. The vibration caused by this traffic load will be superimposed on the structural vibration of the bridge itself, increasing the complexity of the vibration signal. At the same time, when large vehicles such as trucks pass by, violent vibrations will be generated, which may obscure or cover up some of the tiny vibration characteristics of the bridge itself, interfering with the collection and analysis of vibration data. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-channel distributed optical fiber sensing system in order to solve the above problems.

[0004] The present invention provides a multi-channel distributed optical fiber sensing system, comprising:

[0005] A bridge data collection module, which is used to collect bridge data, the bridge data including bridge status data and vehicle vibration data;

[0006] The bridge status data is collected once at each time sampling point, and the vibration, temperature and soundprint data of the key parts of the bridge are obtained through the distributed optical fiber sensing system; the vehicle vibration data is collected in the simulation environment, and the current bridge in a healthy state is modeled in the simulation environment, and the process of vehicle passing is simulated;

[0007] The key parts of the bridge include pile foundation, bearing, beam, expansion joint and cap beam; the time sampling point refers to the moment when data is collected at a predetermined time interval;

[0008] A two-dimensional structural data generation module generates two-dimensional structural data from the collected bridge data. The two-dimensional structural data includes a data matrix and a relationship matrix. One unit of the data matrix represents one-dimensional structural data of an independent object. The independent object includes a pile foundation, a support, a beam, an expansion joint, a cap beam and a vehicle vibration. It represents that one unit of the data matrix only contains the bridge data of the independent object it represents.

[0009] The element in the i-th row and j-th column of the relationship matrix represents the association between the i-th unit of the data matrix and the independent object represented by the j-th unit. If there is an association, the value of this element of the relationship matrix is ​​1, otherwise it is 0;

[0010] The one-dimensional structure data No. 1 includes n data items sorted by time, and the t-th data item represents the independent object data collected at the t-th moment;

[0011] A health status judgment module, which inputs the two-dimensional structure data into a judgment model, wherein the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer;

[0012] The one-dimensional structure data No. 1 is input into the first hidden layer, and the first hidden state is output to the first feature fusion layer. The first feature fusion layer also inputs the fifth hidden state, and the first feature fusion layer outputs the sixth hidden state to the second hidden layer. The second hidden layer also inputs the relationship matrix, and the second hidden layer outputs the second hidden state to the first output layer. The first output layer outputs a health vector, and a component of the health vector represents the health status of a key part of the bridge.

[0013] Furthermore, physical simulation software is used to simulate the current bridge and vehicle in a complete and healthy state according to the design data of the current bridge, and vibration data of the contact point between the vehicle and the bridge is collected in the simulation system.

[0014] Furthermore, the pile foundation and the cap beam are associated with each other in that: the pile foundation and the cap beam are connected;

[0015] The cap beam and the support are related in that: the cap beam is connected to the support;

[0016] The association between the support and the beam means that: the support is connected to the beam;

[0017] The association between beams and expansion joints means that: the beams are connected to the expansion joints;

[0018] The beam is associated with the vehicle vibration when: the vehicle is passing over the beam;

[0019] The existence of a correlation between the expansion joint and the vehicle vibration means that the vehicle is passing through the expansion joint.

[0020] Furthermore, the third hidden layer, the fourth hidden layer and the second feature fusion layer are connected to the second output layer for independent pre-training.

[0021] Further, the static vibration data of the bridge with vehicles are sorted to obtain the second one-dimensional structure data; the second one-dimensional structure data includes n data items sorted by time, and the tth data item represents the static vibration data of the key part of the bridge with vehicles collected at the tth moment;

[0022] The vehicle vibration data is sorted to obtain the third one-dimensional structure data; the third one-dimensional structure data includes n data items sorted by time, and the tth data item represents the vehicle vibration data collected at the tth time;

[0023] The one-dimensional structure data No. 2 is input into the third hidden layer, and the third hidden state is output to the second feature fusion layer. The one-dimensional structure data No. 3 is input into the fourth hidden layer, and the fourth hidden state is output to the second feature fusion layer. The second feature fusion layer outputs the fifth hidden state to the second output layer. The second output layer outputs a vibration vector. A component of the vibration vector represents the vibration data of a key part of the bridge at that moment, including vehicle vibration noise.

[0024] Furthermore, the formula for the first hidden layer is as follows:

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, , and Respectively represent the first, second, and third intermediate states, Represents the tth data item of the one-dimensional structure data. and Represent the t-th and t-1-th first hidden states respectively, n≥t≥1, n represents the total number of data items of the one-dimensional structure data, and when t=1 , tanh is the hyperbolic tangent function, Represents the sigmoid function;

[0030] The formula for the third hidden layer is as follows:

[0031] ;

[0032] in represents the tth third hidden state of the output of the third hidden layer, represents the t-1th third hidden state, Represents the tth data item of the second one-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter;

[0033] The formula for the fourth hidden layer is as follows:

[0034] ;

[0035] in represents the tth fourth hidden state of the output of the fourth hidden layer, represents the t-1th fourth hidden state, Represents the tth data item of the three-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter;

[0036] The formula for the second feature fusion layer is as follows:

[0037] ;

[0038] in represents the fifth hidden state of the output of the second feature fusion layer, Indicates A third hidden state, Indicates a fourth hidden state, n represents the total number of data items of the second one-dimensional structure data and the third one-dimensional structure data;

[0039] The formula for the first feature fusion layer is as follows:

[0040] ;

[0041] in represents the sixth hidden state output by the first feature fusion layer, Indicates The first hidden state;

[0042] The formula for the second hidden layer is as follows:

[0043] ;

[0044] in represents the second hidden state of the vth cell of the data matrix, Represents the data matrix The aggregation coefficient of each unit, Represents the first The set of cells of the data matrix with associated cells, represents the activation function, represents the eighth weight coefficient, Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state;

[0045] ;

[0046] ;

[0047] ;

[0048] in Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state; represents the aggregation weight coefficient, represents the splicing weight coefficient, T represents transposition, represents an exponential function with a natural constant as the base, and LeakyRelu represents the rectified linear unit function.

[0049] Furthermore, the formula of the first output layer is as follows:

[0050] ;

[0051] in Represents the health vector. The cth component of the health vector represents the probability value of the key part of the bridge being healthy. If it is greater than 0.5, it means that the key part of the bridge is healthy. Otherwise, it means that the key part of the bridge is unhealthy. represents the second hidden state of the vth cell of the data matrix, represents the set of cells of the data matrix associated with the vth bridge key part, is the ninth weight parameter, is the sixth bias parameter;

[0052] The calculation formula of the second output layer is as follows:

[0053] ;

[0054] in, represents the vibration vector, and its dth component value represents the vibration data of a key part of the bridge at the dth moment, including the vehicle vibration noise. is the tenth weight parameter, is the seventh bias parameter.

[0055] Furthermore, the loss function of the model is determined for:

[0056] ;

[0057] in, The predicted loss for the health state, The loss predicted for the vibration data, and To balance the hyperparameters of the two losses, the default values ​​are 0.6 and 0.4 respectively; It is expressed as:

[0058] ;

[0059] in represents the total number of training samples, It is The true health status label of the training sample is 0 or 1, 1 means that the key part of the bridge of the training sample is healthy in the experimental environment, and 0 means that the key part of the bridge is healthy or unhealthy in the experimental environment. It is the first The classification labels of training samples, It represents the logarithmic function with the natural constant e as the base; It is expressed as:

[0060] ;

[0061] in represents the total number of training samples, Indicates the number of time sampling points, Indicates Time sampling point Predicted vibration data of key parts of bridges, Indicates Time sampling point Real vibration data of key parts of the bridge.

[0062] The present invention provides a multi-channel distributed optical fiber sensing method, which performs the following steps based on the aforementioned multi-channel distributed optical fiber sensing system:

[0063] Step 301, collecting bridge data, the bridge data including bridge status data and vehicle vibration data;

[0064] Step 302, generating two-dimensional structural data from the collected bridge data, the two-dimensional structural data including a data matrix and a relationship matrix, one unit of the data matrix represents one-dimensional structural data of an independent object, indicating that one unit of the data matrix only contains the bridge data of the independent object represented by it;

[0065] Step 303, input the two-dimensional structure data into the judgment model, the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer. The first output layer outputs a health vector, and a component of the health vector represents the health status of a key part of the bridge.

[0066] The present invention provides a storage medium storing non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the steps of the aforementioned multi-channel distributed optical fiber sensing method can be executed.

[0067] The beneficial effects of the present invention are as follows: by regularly collecting bridge data, the present invention can obtain and analyze the vibration, temperature and soundprint data of key parts of the bridge, fully considering the connection status between the key parts of the bridge and the vibration and noise interference introduced by traffic loads. Compared with determining the health status of the bridge through a single part, the system can simultaneously consider the historical status and dependency relationships of each key part of the bridge, and comprehensively reflect the health status of the entire bridge. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a system module diagram of the present invention;

[0069] Figure 2 It is a flow chart of a multi-channel distributed optical fiber sensing method of the present invention.

[0070] In the figure: a bridge data collection module 101, a two-dimensional structure data generation module 102, and a health status judgment module 103. DETAILED DESCRIPTION

[0071] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0072] In a multi-channel distributed optical fiber sensing system of the present invention, as Figure 1 As shown, including:

[0073] A bridge data collection module 101 is used to collect bridge data, the bridge data including bridge status data and vehicle vibration data;

[0074] The bridge status data is collected once at each time sampling point, and the vibration, temperature and soundprint data of the key parts of the bridge are obtained through the distributed optical fiber sensing system; the vehicle vibration data is collected in the simulation environment, and the current bridge in a healthy state is modeled in the simulation environment, and the process of vehicle passing is simulated;

[0075] The key parts of the bridge include pile foundation, bearings, beams, expansion joints and cap beams;

[0076] In one embodiment of the present invention, a distributed fiber optic sensing system has been deployed at key locations of the bridge to continuously acquire vibration, temperature and soundprint data; at the same time, a traffic flow monitoring system has been installed to obtain vehicle passing information.

[0077] In one embodiment of the present invention, a physical simulation software is used to simulate the current bridge and vehicle in a healthy state in accordance with the design data of the current bridge, and vibration data of the contact point between the vehicle and the bridge is collected in the simulation system (because the vibration of the bridge in an unhealthy state will also affect the collection of vehicle vibration data, so the vehicle vibration data is collected in the simulation system to obtain data close to the vibration noise generated by the actual vehicle);

[0078] In one embodiment of the present invention, during measurement, vehicles are manually planned to pass through the bridge, and a test vehicle is used instead of the actual passing vehicle. The complexity of the test vehicle modeling is lower than that of the actual passing vehicle, so as to reduce the error of the simulation modeling;

[0079] The time sampling point refers to the moment when data is collected at a predetermined time interval;

[0080] In one embodiment of the present invention, the default value of the predetermined time interval is 1 second.

[0081] A two-dimensional structure data generating module 102 generates two-dimensional structure data from the collected bridge data, wherein the two-dimensional structure data includes a data matrix and a relationship matrix, wherein one unit of the data matrix represents one-dimensional structure data of an independent object, wherein the independent objects include pile foundations, bearings, beams, expansion joints, cap beams and vehicle vibrations, wherein one unit of the data matrix representing pile foundations only includes bridge status data of the pile foundations, one unit of the data matrix representing bearings only includes bridge status data of the bearings, one unit of the data matrix representing beams only includes bridge status data of the beams, one unit of the data matrix representing cap beams only includes bridge status data of the cap beams, one unit of the data matrix representing expansion joints only includes bridge status data of the expansion joints, and one unit of the data matrix representing vehicle vibrations only includes vehicle vibration data of the vehicle;

[0082] The element in the i-th row and j-th column of the relationship matrix represents the association between the i-th unit of the data matrix and the independent object represented by the j-th unit. If there is an association, the value of this element of the relationship matrix is ​​1, otherwise it is 0;

[0083] The relationship between the pile foundation and the cap beam means that the pile foundation is connected to the cap beam;

[0084] The cap beam and the support are related in that: the cap beam is connected to the support;

[0085] The association between the support and the beam means that: the support is connected to the beam;

[0086] The association between beams and expansion joints means that: the beams are connected to the expansion joints;

[0087] The beam is associated with the vehicle vibration when: the vehicle is passing over the beam;

[0088] The relationship between expansion joints and vehicle vibrations refers to: the vehicle is passing through the expansion joint;

[0089] The one-dimensional structure data No. 1 includes n data items sorted by time, and the t-th data item represents the independent object data collected at the t-th moment;

[0090] The health status judgment module 103 inputs the two-dimensional structure data into a judgment model, wherein the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer;

[0091] Inputting the one-dimensional structure data No. 1 into the first hidden layer, outputting the first hidden state to the first feature fusion layer, the first feature fusion layer also inputs the fifth hidden state, the first feature fusion layer outputs the sixth hidden state to the second hidden layer, the second hidden layer also inputs the relationship matrix, the second hidden layer outputs the second hidden state to the first output layer, the first output layer outputs a health vector, a component of the health vector represents the health state of a key part of the bridge;

[0092] The third hidden layer, the fourth hidden layer and the second feature fusion layer are connected to the second output layer for independent pre-training. During the independent pre-training, the vibration of the vehicle and the independent vibration of the bridge are observed, and the process of the vibration of the vehicle being transmitted to the key nodes of the bridge to generate the vibration of the key nodes of the bridge after being affected is fitted. The correlation relationship between the transmission of the vibration of the vehicle to the key nodes of the bridge can be pattern recognized, and the correlation relationship can be used to remove the mixed vibration of the vehicle in the measured vibration of the key nodes of the bridge;

[0093] The static vibration data of the bridge with vehicles are sorted to obtain the second one-dimensional structure data; the second one-dimensional structure data includes n data items sorted by time, and the tth data item represents the static vibration data of the key part of the bridge with vehicles collected at the tth time;

[0094] The vehicle vibration data is sorted to obtain the third one-dimensional structure data; the third one-dimensional structure data includes n data items sorted by time, and the tth data item represents the vehicle vibration data collected at the tth time;

[0095] Input the second one-dimensional structure data into the third hidden layer, output the third hidden state to the second feature fusion layer, input the third one-dimensional structure data into the fourth hidden layer, output the fourth hidden state to the second feature fusion layer, the second feature fusion layer outputs the fifth hidden state to the second output layer, and the second output layer outputs a vibration vector, wherein a component of the vibration vector represents vibration data of a key part of the bridge at that moment including vehicle vibration noise;

[0096] In one embodiment of the present invention, static bridge vibration data with a vehicle, vehicle vibration data and vibration vectors are all obtained through simulation, wherein the static bridge vibration with a vehicle is equivalent to pure bridge vibration.

[0097] In one embodiment of the present invention, the formula of the first hidden layer is as follows:

[0098]

[0099]

[0100]

[0101]

[0102] in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, , and Respectively represent the first, second, and third intermediate states, Represents the tth data item of the one-dimensional structure data. and Represent the t-th and t-1-th first hidden states respectively, n≥t≥1, n represents the total number of data items of the one-dimensional structure data, and when t=1 , tanh is the hyperbolic tangent function, Represents the sigmoid function.

[0103] In one embodiment of the present invention, the formula of the third hidden layer is as follows:

[0104]

[0105] in represents the tth third hidden state of the output of the third hidden layer, represents the t-1th third hidden state, Represents the tth data item of the second one-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter.

[0106] In one embodiment of the present invention, the formula of the fourth hidden layer is as follows:

[0107]

[0108] in represents the tth fourth hidden state of the output of the fourth hidden layer, represents the t-1th fourth hidden state, Represents the tth data item of the three-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter.

[0109] In one embodiment of the present invention, the formula of the second feature fusion layer is as follows:

[0110]

[0111] in represents the fifth hidden state of the output of the second feature fusion layer, Indicates A third hidden state, Indicates A fourth hidden state, n represents the total number of data items of the second one-dimensional structure data and the third one-dimensional structure data.

[0112] In one embodiment of the present invention, the formula of the first feature fusion layer is as follows:

[0113]

[0114] in represents the sixth hidden state output by the first feature fusion layer, Indicates The first hidden state.

[0115] In one embodiment of the present invention, the formula of the second hidden layer is as follows:

[0116]

[0117] in represents the second hidden state of the vth cell of the data matrix, Represents the data matrix The aggregation coefficient of each unit, Represents the first The set of cells of the data matrix with associated cells, represents the activation function, represents the eighth weight coefficient, Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state;

[0118]

[0119]

[0120]

[0121] in Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state; represents the aggregation weight coefficient, represents the splicing weight coefficient, T represents transposition, represents an exponential function with a natural constant as the base, and LeakyRelu represents the rectified linear unit function.

[0122] In one embodiment of the present invention, the formula of the first output layer is as follows:

[0123]

[0124] in Represents the health vector. The cth component of the health vector represents the probability value of the key part of the bridge being healthy. If it is greater than 0.5, it means that the key part of the bridge is healthy. Otherwise, it means that the key part of the bridge is unhealthy. represents the second hidden state of the vth cell of the data matrix, represents the set of cells of the data matrix associated with the vth bridge key part, is the ninth weight parameter, is the sixth bias parameter.

[0125] In one embodiment of the present invention, the calculation formula of the second output layer is as follows:

[0126]

[0127] in, represents the vibration vector, and its dth component value represents the vibration data of a key part of the bridge at the dth moment, including the vehicle vibration noise. is the tenth weight parameter, is the seventh bias parameter.

[0128] In one embodiment of the present invention, the loss function of the judgment model is for:

[0129]

[0130] in, The predicted loss for the health state, The loss predicted for the vibration data, and To balance the hyperparameters of the two losses, the default values ​​are 0.6 and 0.4 respectively;

[0131] It is expressed as:

[0132]

[0133] in represents the total number of training samples, It is The true health status label of the training sample is 0 or 1, 1 means that the key part of the bridge of the training sample is healthy in the experimental environment, and 0 means that the key part of the bridge is healthy or unhealthy in the experimental environment. It is the first The classification labels of training samples, Represents the logarithmic function with the natural constant e as the base.

[0134] It is expressed as:

[0135]

[0136] in represents the total number of training samples, Indicates the number of time sampling points, Indicates Time sampling point Predicted vibration data of key parts of bridges, Indicates Time sampling point Real vibration data of key parts of the bridge.

[0137] In one embodiment of the present invention, a multi-channel distributed optical fiber sensing method is provided, such as Figure 2 As shown, the following steps are included:

[0138] Step 301, collecting bridge data, the bridge data including bridge status data and vehicle vibration data;

[0139] Step 302, generating two-dimensional structural data from the collected bridge data, the two-dimensional structural data including a data matrix and a relationship matrix, one unit of the data matrix represents one-dimensional structural data of an independent object, indicating that one unit of the data matrix only contains the bridge data of the independent object represented by it;

[0140] Step 303, input the two-dimensional structure data into the judgment model, the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer. The first output layer outputs a health vector, and a component of the health vector represents the health status of a key part of the bridge.

[0141] At least one embodiment of the present invention provides a storage medium storing non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the steps of the aforementioned multi-channel distributed optical fiber sensing method can be executed.

[0142] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation mode. The above-mentioned specific implementation mode is merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make more forms of equivalent embodiments, all of which are within the protection of this embodiment.

Claims

1. A multi-channel distributed optical fiber sensing system, characterized in that: include: A bridge data collection module, which is used to collect bridge data, the bridge data including bridge status data and vehicle vibration data; The bridge status data is collected once at each time sampling point, and the vibration, temperature and soundprint data of the key parts of the bridge are obtained through the distributed optical fiber sensing system; the vehicle vibration data is collected in the simulation environment, and the current bridge in a healthy state is modeled in the simulation environment, and the process of vehicle passing is simulated; The key parts of the bridge include pile foundation, bearing, beam, expansion joint and cap beam; the time sampling point refers to the moment when data is collected at a predetermined time interval; A two-dimensional structural data generation module generates two-dimensional structural data from the collected bridge data. The two-dimensional structural data includes a data matrix and a relationship matrix. One unit of the data matrix represents one-dimensional structural data of an independent object. The independent object includes a pile foundation, a support, a beam, an expansion joint, a cap beam and a vehicle vibration. It represents that one unit of the data matrix only contains the bridge data of the independent object it represents. The element in the i-th row and j-th column of the relationship matrix represents the association between the i-th unit of the data matrix and the independent object represented by the j-th unit. If there is an association, the value of this element of the relationship matrix is ​​1, otherwise it is 0; The one-dimensional structure data No. 1 includes n data items sorted by time, and the t-th data item represents the independent object data collected at the t-th moment; A health status judgment module, which inputs the two-dimensional structure data into a judgment model, wherein the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer; The one-dimensional structure data No. 1 is input into the first hidden layer, and the first hidden state is output to the first feature fusion layer. The first feature fusion layer also inputs the fifth hidden state, and the first feature fusion layer outputs the sixth hidden state to the second hidden layer. The second hidden layer also inputs the relationship matrix, and the second hidden layer outputs the second hidden state to the first output layer. The first output layer outputs a health vector, and a component of the health vector represents the health status of a key part of the bridge.

2. A multi-channel distributed optical fiber sensing system according to claim 1, characterized in that: Physical simulation software is used to simulate the current bridge and vehicle in a complete and healthy state according to the design data of the current bridge, and vibration data of the contact point between the vehicle and the bridge is collected in the simulation system.

3. A multi-channel distributed optical fiber sensing system according to claim 1, characterized in that: The relationship between the pile foundation and the cap beam means that the pile foundation is connected to the cap beam; The cap beam and the support are related in that: the cap beam is connected to the support; The association between the support and the beam means that: the support is connected to the beam; The association between beams and expansion joints means that: the beams are connected to the expansion joints; The beam is associated with the vehicle vibration when: the vehicle is passing over the beam; The existence of a correlation between the expansion joint and the vehicle vibration means that the vehicle is passing through the expansion joint.

4. A multi-channel distributed optical fiber sensing system according to claim 1, characterized in that: The third hidden layer, the fourth hidden layer, and the second feature fusion layer are connected to the second output layer for independent pre-training.

5. A multi-channel distributed optical fiber sensing system according to claim 4, characterized in that: The static vibration data of the bridge with vehicles are sorted to obtain the second one-dimensional structure data; the second one-dimensional structure data includes n data items sorted by time, and the tth data item represents the static vibration data of the key part of the bridge with vehicles collected at the tth time; The vehicle vibration data is sorted to obtain the third one-dimensional structure data; the third one-dimensional structure data includes n data items sorted by time, and the tth data item represents the vehicle vibration data collected at the tth time; The one-dimensional structure data No. 2 is input into the third hidden layer, and the third hidden state is output to the second feature fusion layer. The one-dimensional structure data No. 3 is input into the fourth hidden layer, and the fourth hidden state is output to the second feature fusion layer. The second feature fusion layer outputs the fifth hidden state to the second output layer. The second output layer outputs a vibration vector. A component of the vibration vector represents the vibration data of a key part of the bridge at that moment, including vehicle vibration noise.

6. A multi-channel distributed optical fiber sensing system according to claim 5, characterized in that: The formula for the first hidden layer is as follows: ； ； ； ； in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, , and Respectively represent the first, second, and third intermediate states, Represents the tth data item of the one-dimensional structure data. and Represent the t-th and t-1-th first hidden states respectively, n≥t≥1, n represents the total number of data items of the one-dimensional structure data, and when t=1 , tanh is the hyperbolic tangent function, Represents the sigmoid function; The formula for the third hidden layer is as follows: ； in represents the tth third hidden state of the output of the third hidden layer, represents the t-1th third hidden state, Represents the tth data item of the second one-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter; The formula for the fourth hidden layer is as follows: ； in represents the tth fourth hidden state of the output of the fourth hidden layer, represents the t-1th fourth hidden state, Represents the tth data item of the three-dimensional structure data, and are the fourth and fifth weight parameters, is the fourth bias parameter; The formula for the second feature fusion layer is as follows: ； in represents the fifth hidden state of the output of the second feature fusion layer, Indicates A third hidden state, Indicates a fourth hidden state, n represents the total number of data items of the second one-dimensional structure data and the third one-dimensional structure data; The formula for the first feature fusion layer is as follows: ； in represents the sixth hidden state output by the first feature fusion layer, Indicates The first hidden state; The formula for the second hidden layer is as follows: ； in represents the second hidden state of the vth cell of the data matrix, Represents the data matrix The aggregation coefficient of each unit, Represents the first The set of cells of the data matrix with associated cells, represents the activation function, represents the eighth weight coefficient, Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state; ； ； ； in Represents the data matrix When the one-dimensional structure data of the first unit is input into the first hidden layer, the output A sixth hidden state obtained by summing the first hidden state and the corresponding fifth hidden state; represents the aggregation weight coefficient, represents the splicing weight coefficient, T represents transposition, represents an exponential function with a natural constant as the base, and LeakyRelu represents the rectified linear unit function.

7. A multi-channel distributed optical fiber sensing system according to claim 6, characterized in that: The formula for the first output layer is as follows: ； in Represents the health vector. The cth component of the health vector represents the probability value of the key part of the bridge being healthy. If it is greater than 0.5, it means that the key part of the bridge is healthy. Otherwise, it means that the key part of the bridge is unhealthy. represents the second hidden state of the vth cell of the data matrix, represents the set of cells of the data matrix associated with the vth bridge key part, is the ninth weight parameter, is the sixth bias parameter; The calculation formula of the second output layer is as follows: ； in, represents the vibration vector, and its dth component value represents the vibration data of a key part of the bridge at the dth moment, including the vehicle vibration noise. is the tenth weight parameter, is the seventh bias parameter.

8. The multi-channel distributed optical fiber sensing system according to claim 1, characterized in that: Determine the loss function of the model for: ； in, The predicted loss for the health state, The loss predicted for the vibration data, and To balance the hyperparameters of the two losses, the default values ​​are 0.6 and 0.4 respectively; It is expressed as: ； in represents the total number of training samples, It is The true health status label of the training sample is 0 or 1, where 1 means that the key part of the bridge of the training sample is healthy in the experimental environment, and 0 means that the key part of the bridge is healthy or unhealthy in the experimental environment. It is the first The classification labels of training samples, It represents the logarithmic function with the natural constant e as the base; It is expressed as: ； in represents the total number of training samples, Indicates the number of time sampling points, Indicates Time sampling point Predicted vibration data of key parts of bridges, Indicates Time sampling point Real vibration data of key parts of the bridge.

9. A multi-channel distributed optical fiber sensing method, characterized in that: Based on a multi-channel distributed optical fiber sensing system as described in any one of claims 1 to 8, the following steps are performed: Step 301, collecting bridge data, the bridge data including bridge status data and vehicle vibration data; Step 302, generating two-dimensional structural data from the collected bridge data, the two-dimensional structural data including a data matrix and a relationship matrix, one unit of the data matrix represents one-dimensional structural data of an independent object, indicating that one unit of the data matrix only contains the bridge data of the independent object represented by it; Step 303, input the two-dimensional structure data into the judgment model, the judgment model includes a first hidden layer, a first feature fusion layer, a second hidden layer, a third hidden layer, a fourth hidden layer, a second feature fusion layer, a first output layer, and a second output layer. The first output layer outputs a health vector, and a component of the health vector represents the health status of a key part of the bridge.

10. A storage medium, characterized in that: It stores non-transitory computer-readable instructions, and when the non-transitory computer-readable instructions are executed by a computer, the steps of the multi-channel distributed optical fiber sensing method as claimed in claim 9 can be executed.

Images