Large-span railway bridge beam end area state detecting and monitoring system

Abstract

The invention discloses a long-span railway bridge beam end area state monitoring system which comprises a beam end environment monitoring module, a beam surface state monitoring module, a pier beam state monitoring module and a network transmission and power supply module. The long-span railway bridge beam end area state detecting and monitoring system can be applied to beam ends of different bridge structures such as a long-span railway continuous beam bridge, an arch bridge, a cable-stayed bridge and a suspension bridge, and the accuracy, efficiency and safety of bridge beam end state health management can be remarkably improved. Through the functions of real-time monitoring, disease recognition, data analysis, fault early warning and the like, deterioration of bridge diseases can be effectively prevented, the service life of a bridge is prolonged, the maintenance cost is reduced, smooth operation of railway traffic is ensured, and along with continuous development of the intelligent technology, the detection and monitoring system and method are wide in application prospect and high in practicability. The method plays an important role in the field of intelligent transportation in the future, and promotes railway bridge management to be in a more scientific and refined management mode.

Description

Technical Field

[0001] The present invention relates to the technical field of inspection and monitoring of the beam end area of long-span high-speed railway bridges, and particularly relates to a state inspection and monitoring system for the beam end area of long-span railway bridges. Background Art

[0002] The scale of railway construction in China has developed rapidly. Most long-span railway bridges are control projects for crossing rivers and seas and connecting important economic circle lines. High-speed trains have characteristics such as high running speed and high smoothness requirements. During the service period of long-span high-speed railway bridges, the time-varying characteristics of the line-bridge system are significant under the action of complex loads such as temperature, wind, and trains. Especially in the long-span beam end area, the smoothness and stability are poor. Diseases such as concrete corrosion and cracking at the beam end, uneven settlement of piers, expansion and contraction of bearings, skew of steel sleepers, and jamming of scissors forks lead to speed limits for trains on long-span high-speed railway bridges. Accurately monitoring the state of the long-span beam end area during the operation period is crucial for ensuring train operation safety. The current common inspection and monitoring means for the beam end area are as follows: ① Manual regular inspection, through applying for a skylight to conduct on-site inspections, using measuring tools and taking photos to check the specific index status of the railway beam end area. The skylight time is generally 2-3 hours, which is short and discontinuous, and the inspection efficiency is low; ② Deploying contact sensor devices separately at the expansion joints of the track beam ends, and monitoring the displacement status of the beam end expansion joints through point-by-point layout. On the one hand, the monitoring indicators of this method are limited, and on the other hand, there are safety hazards such as cable detachment for contact devices, affecting train operation safety; ③ Diseases of each component at the beam end are not discovered in time, and the comprehensive index changes of each component such as the track, bridge, bearing, and pier at the beam end cannot be grasped uniformly, especially the simultaneous effect changes under accidental loads and high-frequency train operation. There is a lack of a method for comprehensively evaluating the state of the beam end area, and overall linkage evaluation has not been achieved. Therefore, there is an urgent need for a state inspection and monitoring system for the beam end area of long-span railway bridges to solve the problems of the existing technology. Summary of the Invention

[0003] In view of the above problems, the present invention is proposed to provide a state inspection and monitoring system for the beam end area of long-span railway bridges that overcomes the above problems or at least partially solves the above problems.

[0004] In order to solve the above technical problems, the embodiments of the present application disclose the following technical solutions:

[0005] In a first aspect, an embodiment of the present invention discloses a state inspection and monitoring system for the beam end area of long-span railway bridges, including: a beam end environment monitoring module, a beam surface state inspection and monitoring module, a pier-beam state monitoring module, a network transmission and power supply module;

[0006] The beam end environment monitoring module is used to monitor the environmental characteristics of the beam end of the long-span railway bridge, and the environmental characteristics specifically include the wind speed, wind direction, environmental temperature, and humidity of the beam end bridge deck;

[0007] The beam surface condition monitoring module is used to monitor various index conditions of the beam surface at the beam end, realize the evaluation of the apparent deterioration characteristics of the beam end area, the evaluation of the condition of the beam end expansion device, and the evaluation of the beam surface displacement at the beam end;

[0008] The pier-beam condition monitoring module is used to monitor the pier top rotation and vibration of the pier and the displacements of the main bridge and the approach bridge, and realize the evaluation of various response indexes of the pier-beam;

[0009] The network transmission and power supply module is used for data transmission and online power supply of the beam end environment monitoring module, the beam surface condition monitoring module and the pier-beam condition monitoring module, and ensures the smooth transmission of data and images and the stable voltage.

[0010] Furthermore, the beam end environment monitoring module includes a wind speed and direction sensor and a temperature and humidity sensor. The wind speed and direction sensor is installed on the cross-arm flange perpendicular to the longitudinal direction of the bridge deck and is used to test the wind speed and direction environment of the beam end area; the temperature and humidity sensor is installed at the connection position of the cross-arm and the vertical pole and is used to test the temperature and humidity environment of the beam end area.

[0011] Furthermore, the beam surface condition monitoring module includes an end deterioration characteristic visual recognition sub-module, a expansion device displacement recognition sub-module, and a beam end GNSS displacement monitoring sub-module; among them:

[0012] The end deterioration characteristic visual recognition sub-module is used to identify the abnormal states of each component at the beam end, extract the apparent disease characteristics of the beam end components through non-contact visual analysis methods. The disease indexes at least include scissor cross abnormality, beam end sleeper breakage, beam end movable sleeper skew, beam end track spalling and chipping, beam end rail light band abnormality, beam end rail corrugation wear, and beam end fastener spring clip fracture; the apparent deterioration characteristics of the beam end area are evaluated through the disease indexes;

[0013] The expansion device displacement recognition sub-module is used to monitor the representative displacement states of each component of the beam end expansion device. The representative displacement states at least include the distance between movable sleepers, the vertical displacement of the steel sleeper, the rail expansion displacement, the longitudinal beam expansion amount, and the scissor cross XZ plane movement track; the condition of the expansion device is evaluated through the representative displacement states of each component;

[0014] The beam end GNSS displacement monitoring sub-module is used to monitor the absolute longitudinal X displacement, absolute transverse Y displacement, and absolute vertical Z displacement of the beam end. The beam end beam surface displacement is evaluated through the absolute longitudinal X displacement, absolute transverse Y displacement, and absolute vertical Z displacement of the beam end.

[0015] Furthermore, the end deterioration characteristic visual recognition sub-module is used to identify the abnormal states of each component at the beam end, extract the apparent disease characteristics of the beam end components through non-contact visual analysis methods. The specific methods include:

[0016] Obtain a dataset of the disease characteristics at the beam ends of railway bridges, and preprocess the dataset. The disease characteristics at least include sleeper breakage, abnormal light band, sleeper spalling, corrugation wear, and fish-scale corrugation.

[0017] Build and train a disease recognition neural network. The recognition neural network model includes an input layer, a convolutional layer, a pooling layer, and a transposed convolutional layer. Among them, the input layer processes the initial input of the neural network and is used to receive the original image data from the monitoring system. The preprocessed data passed through the input layer will enter the subsequent convolutional layer as the basis for feature extraction. The convolutional layer processes the monitoring data through the sliding operation of the convolutional kernel, performs convolutional calculations on the premise of restricting the model complexity and reducing overfitting, extracts signal feature parameters, and the size of the intermediate matrix output after convolutional calculation is as follows:

[0018]

[0019] Where i is the input size collected by the monitoring system, p is the boundary value, s is the step size, and k is the convolutional kernel step size.

[0020] Further, when the 2x2 convolutional kernel in the disease recognition neural network operates on an input matrix with a size of 3x3, taking the 2x2 area in the upper left corner of the input matrix as the starting point, through the product operation and accumulation process between neurons, calculate the first output value b11; then the convolutional kernel moves to the right by a preset step size distance to reach the next 2x2 area, and repeat the same convolutional operation to obtain the output value b12; the operation process continues until the convolutional kernel matrix advances to the end of the input matrix, and finally a 2x2 output feature matrix is formed.

[0021] Further, after the convolutional layer of the disease recognition neural network, max pooling processing is performed. After the pooling layer processes the data, the disease recognition neural network will perform convolution and pooling operations on the data again to further extract image features; after the convolutional calculation is completed, the data is imported into the transposed convolutional layer to map the intermediate data with a small resolution to the analysis image with a large resolution, introduce the deconvolution kernel into the analysis data to calculate and rearrange the data matrix, so as to restore to the original input data size; the calculation formula for the deconvolution size is as follows:

[0022] o' = s(i - 1) + k' - 2p'

[0023] Where i is the size of the intermediate matrix of the monitoring system, p is the boundary value, s is the step size, and k is the deconvolution kernel step size;

[0024] The intermediate matrix after the deconvolution layer expansion will be processed by the fully connected layer, express the prediction result in the form of a probability distribution, and complete the accurate determination of the disease category at the beam end. The calculation formula of the fully connected layer is as follows:

[0025]

[0026] Among them, O is the output matrix of the fully connected layer, I is the intermediate matrix input to the fully connected layer, w is the weight coefficient of the neurons in the corresponding layer, and B represents the bias of the fully connected layer;

[0027] After completing the fully connected processing, the intermediate matrix is analyzed and normalized by the classifier, so that the unnormalized prediction of the intermediate matrix is transformed into non-negative numbers and the sum is 1. The formula for its normalized probability distribution is as follows:

[0028]

[0029] where n is the total number of neurons in this layer, o i represents the log probability of the i-th node;

[0030] The cross-entropy loss is used as the model evaluation quantization evaluation to express the fitting degree calculated by the recognition module. The formula of the cross-entropy judgment function is as follows:

[0031]

[0032] where q(x i ) is the conclusion probability obtained through the disease recognition analysis and processing, and p(x i ) is the actual observed result.

[0033] Furthermore, the disease recognition neural network continuously iterates and updates the data set, uses the steepest descent method to optimize the parameters in the model, analyzes the loss function and accuracy of the neural network model by real-time monitoring the performance of the neural network model, and improves the training accuracy of the model; analyzes the boundary range of the model prediction and the boundary situation of the actual situation in the end as an investigation, and investigates its precision, recall rate, and intersection over union as the evaluation indicators of the model accuracy performance; the evaluation formulas are as follows:

[0034]

[0035] Among them, TP refers to the number of instances correctly identified as positive by the model, reflecting the correct judgment ability of the model for positive classes; TN refers to the number of instances correctly identified as negative by the model, reflecting the correct judgment ability of the model for negative classes; correspondingly, FP refers to the number of negative class instances misjudged as positive by the model; FN refers to the number of positive class instances misjudged as negative by the model;

[0036] Integrate the trained model into the edge computing gateway to process, extract features, and identify damages for the beam end image information collected in real time in the main bridge and approach bridge areas. Integrate and process the identified disease information after processing to form a disease assessment report for evaluating the severity of diseases in the beam end area, predicting the future severity, and providing maintenance suggestions, and generate a monitoring report from the identification results for storage.

[0037] Furthermore, the specific working methods of the expansion device displacement identification sub-module and the beam end GNSS displacement monitoring sub-module include:

[0038] Fix the beam surface inspection monitoring device, and place the measuring point targets on the switch rail, the left longitudinal beam of the rail lifting device, the left scissor cross, the stock rail, the movable sleeper, the right longitudinal beam of the rail lifting device, and the right scissor cross by surface mounting.

[0039] After the beam surface inspection monitoring device is installed and fixed, conduct device equipment debugging, calibrate the initial values of the X, Y, and Z axes of the GNSS Beidou equipment in the beam end GNSS displacement monitoring sub-module, and the change amounts in the X, Y, and Z directions are x 17 , y 17 , z 17 respectively; the first optoelectronic deflection meter is responsible for the target monitoring areas of the left scissor cross, the stock rail, the movable sleeper, and the right scissor cross, and the second optoelectronic deflection meter is responsible for the target monitoring areas of the switch rail, the left longitudinal beam of the rail lifting device, and the right longitudinal beam of the rail lifting device. The two optoelectronic deflection meters record the initial values of the X and Z axes of the targets in their respective responsible areas; correct the actual longitudinal expansion displacement of the switch rail, the actual longitudinal expansion displacement of the left longitudinal beam of the rail lifting device, and the actual longitudinal expansion displacement of the right longitudinal beam of the rail lifting device respectively through the change amount x 17 in the X direction of the Beidou device, and evaluate the target monitoring areas of the left scissor cross, the movable sleeper, and the right scissor cross by observing the change trend of the displacement value data in the X and Z directions.

[0040] Furthermore, the pier and beam state monitoring module is used to monitor the pier top rotation angle and vibration, and the displacements of the main bridge and approach bridge. The specific method includes: by arranging displacement gauges, inclinometers, and vibration gauges, evaluate the change states of the main bridge girder, approach bridge girder, and connecting pier components. The obtained indicators at least include the pier top rotation angle, pier top vibration, main bridge girder displacement, and approach bridge girder displacement; all the indicators involving displacement measurement in the test process are the integrated action amounts in each direction, and each single displacement component has a functional relationship with the change of the pier top displacement. The pier top height is h d , the longitudinal deflection rotation angle of the pier top is θ x , θ x is positive for clockwise deflection and negative for counterclockwise deflection. When the actual pier top has a rotation deviation, both the vertical and longitudinal displacements of the pier top change.

[0041] Furthermore, the network transmission and power supply module includes an integrated processing sub-module, a private network transmission sub-module, and a voltage stabilization power supply sub-module; the integrated processing sub-module synchronously collects data of each module in an integrated manner of a line ring network, and uses an industrial switch for internal ring network communication transmission. When a single device fails, the signal data can continue to be transmitted along the ring link, realizing the function that the damage of a single component does not affect the overall operation of the device; the private network transmission sub-module and the voltage stabilization power supply sub-module transmit and supply power to the test data of each module through an edge computing gateway.

[0042] The beneficial effects of the above technical solutions provided by the embodiments of the present invention at least include:

[0043] The present invention discloses a state detection and monitoring system for the beam end area of a long-span railway bridge, including: a beam end environment monitoring module, a beam surface state detection and monitoring module, a pier-beam state monitoring module, and a network transmission and power supply module; the beam end environment monitoring module is used to monitor the environmental characteristics of the beam end of the long-span railway bridge, and the environmental characteristics specifically include the wind speed, wind direction, environmental temperature, and humidity of the beam end bridge deck; the beam surface state detection and monitoring module is used to detect and monitor various index states of the beam end beam surface, realize the evaluation of the apparent deterioration characteristics of the beam end area, the evaluation of the state of the beam end expansion device, and the evaluation of the displacement of the beam end beam surface; the pier-beam state monitoring module is used to monitor the rotation angle and vibration of the pier top and the displacement of the main bridge and the approach bridge, and realize the evaluation of each response index of the pier-beam; the network transmission and power supply module is used to transmit data and provide on-line power supply for the beam end environment monitoring module, the beam surface state detection and monitoring module, and the pier-beam state monitoring module, ensuring smooth data and image transmission and stable voltage.

[0044] A state detection and monitoring system for the beam end area of a long-span railway bridge disclosed by the present invention can be applied to the beam ends of different bridge structures such as continuous beam bridges, arch bridges, cable-stayed bridges, and suspension bridges of long-span railway bridges, and can significantly improve the accuracy, efficiency, and safety of the state health management of the bridge beam end. Through functions such as real-time monitoring, disease identification, data analysis, and fault warning, it can effectively prevent the deterioration of bridge diseases, extend the service life of the bridge, reduce maintenance costs, and ensure the smooth operation of railway traffic. With the continuous development of intelligent technologies, the application prospects of the detection and monitoring system and method of the present invention are broad, and it will play an important role in the field of intelligent transportation in the future, promoting the management of railway bridges towards a more scientific and refined management mode.

[0045] The following will further describe the technical solutions of the present invention in detail through the drawings and embodiments. Brief Description of the Drawings

[0046] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the accompanying drawings:

[0047] Figure 1 It is a structural diagram of a state inspection and monitoring system for the beam end area of a long-span railway bridge in Embodiment 1 of the present invention;

[0048] Figure 2 It is an implementation schematic diagram of a state monitoring system for the beam end area of a long-span railway bridge in Embodiment 1 of the present invention;

[0049] Figure 3 It is a longitudinal sectional schematic diagram of the beam end area inspection and monitoring system device in Embodiment 1 of the present invention;

[0050] Figure 4 It is a beam surface inspection and monitoring device and implementation schematic diagram in Embodiment 1 of the present invention;

[0051] Figure 5 It is a specific architecture schematic diagram of the disease identification sub-module in Embodiment 1 of the present invention;

[0052] Figure 6 It is a schematic diagram of the convolution operation process in Embodiment 1 of the present invention;

[0053] Figure 7 It is a sub-schematic diagram of the pier-beam state monitoring module in Embodiment 1 of the present invention;

[0054] Figure 8 It is a schematic diagram for analyzing the displacement of the pier top deviation in Embodiment 1 of the present invention. Detailed implementation manners

[0055] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0056] In order to solve the problems existing in the prior art, the embodiments of the present invention provide a state inspection and monitoring system for the beam end area of a long-span railway bridge.

[0057] Embodiment 1

[0058] The embodiments of the present invention disclose a state inspection and monitoring system for the beam end area of a long-span railway bridge, as shown in Figure 1 and 2 , including: a beam end environment monitoring module, a beam surface state inspection and monitoring module, a pier-beam state monitoring module, a network transmission and power supply module;

[0059] The beam-end environment monitoring module is used to monitor the environmental characteristics at the beam ends of long-span railway bridges. The environmental characteristics specifically include the wind speed, wind direction, environmental temperature, and humidity of the bridge deck at the beam ends.

[0060] In this embodiment, the beam-end environment monitoring module includes a wind speed and direction sensor and a temperature and humidity sensor. The wind speed and direction sensor is installed on the cross-arm flange perpendicular to the longitudinal direction of the bridge deck to test the wind speed and direction environmental conditions in the beam-end area; the temperature and humidity sensor is installed at the connection position of the cross-arm and the vertical pole to test the temperature and humidity environmental conditions in the beam-end area.

[0061] To better understand this embodiment, the monitoring devices involved in the system are described in detail. The overall longitudinal section layout of the monitoring devices on-site is as Figure 3 shown.

[0062] The explanations for each label are as follows: 1 - Beam surface monitoring device, 2 - Approach bridge main beam, 3 - Edge gateway local box, 4 - Transmission optical fiber, 5 - Power supply cable, 6 - Approach bridge main beam displacement gauge, 7 - Integrated bridge tray, 8 - Bridge pier, 9 - Pier top inclinometer and vibration meter, 10 - Main bridge displacement gauge, 11 - Beam joint displacement gauge, 12 - Main bridge main beam. The entire system realizes the monitoring and early warning of the state monitoring and state assessment in the beam-end area through the integrated beam-end environment monitoring module, beam surface state monitoring module, pier-beam state monitoring module, network transmission, and power supply module.

[0063] The devices involved in the monitoring system are further described. First, it is the beam-end environment monitoring module and the beam surface state monitoring module, corresponding to 1 - Beam surface monitoring device and each monitoring index of the bridge deck. The specific devices and implementation diagrams are as Figure 4 shown, and the following is an expanded description.

[0064] The explanations for each label of 1 - Beam surface monitoring device are as follows: 13 - Wind speed and direction sensor, 14 - Hexagon socket head cap screw M6*16, 15 - Wind speed and direction cross-arm, 16 - Temperature and humidity sensor, 17 - GNSS Beidou, 18 - High-definition camera (main bridge area), 19 - Photoelectric deflection gauge (main bridge area), 20 - Photoelectric deflection gauge (approach bridge area), 21 - High-definition camera (approach bridge area), 22 - Hexagon head bolt M10*250, 23 - Clamp (φ80), 24 - Clamp (φ114), 25 - Guardrail, 26 - Hexagon head bolt M20*70, 27 - Steel backing plate, 28 - Concrete base.

[0065] The beam surface inspection and monitoring device is welded to the steel bridge deck through steel backing plates. The steel backing plates are processed with single-sided bevel grooves. During installation, the welding quality needs to be ensured. The vertical pole is connected to the steel backing plate through hexagon socket head cap screws. To ensure the stability of the vertical pole, a plain concrete foundation with a length, width, and height of 400*400*300 mm is poured at the vertical pole foundation to ensure the stability and reliability of the bridge deck inspection and monitoring device.

[0066] The welded parts of each device component are uniformly treated with spray painting for rust prevention. Especially for the large-span bridge structures in complex environments, affected by the environment, various steel components need to focus on improving the rust and corrosion prevention level to ensure the durability of the structure.

[0067] Large-span railway bridges are generally located in complex environments. Especially for the influence of extreme weather such as typhoons, an additional lateral connection protection measure is added. The vertical pole is fixed to the bridge railing in the form of screws and clamps. A total of 4 hexagon socket head cap screws are used to connect the railing and the vertical pole clamp to prevent overturning under the action of extreme crosswinds and affect the operation safety of the train. Three bracket structures are designed on the main pole of the vertical pole. Each bracket structure is equipped with a wind speed and direction sensor and two optoelectronic and high-definition camera integrated machines. The wind speed and direction sensor is installed on the cross-arm flange perpendicular to the longitudinal direction of the bridge deck to measure the crosswind environment in the beam end area. The temperature and humidity sensor is installed at the connection position between the cross-arm and the vertical pole. The other two brackets are respectively equipped with optoelectronic and high-definition camera integrated machines, which are responsible for inspecting and monitoring various indicators of the main bridge and approach bridge. The GNSS Beidou device is fixed on the top of the vertical pole by means of flange connection.

[0068] In this embodiment, the wind speed, direction, temperature, and humidity sensors in the beam surface inspection and monitoring device realize the function of the beam end environment monitoring module. The sampling frequency of each sensor is 1 Hz. The real-time data is transmitted through the signal line and exits from the bottom of the vertical pole inside the rod and is connected to the edge gateway local box located inside the box girder to monitor the environmental characteristics of the bridge beam end and provide basic environmental data support.

[0069] The beam surface state inspection and monitoring module is used to inspect and monitor various index states of the beam end beam surface, realize the evaluation of the apparent deterioration characteristics of the beam end area, the evaluation of the state of the beam end expansion device, and the evaluation of the displacement of the beam end beam surface;

[0070] In this embodiment, the beam surface state inspection and monitoring module includes an end deterioration characteristic visual recognition sub-module, an expansion device displacement recognition sub-module, and a beam end GNSS displacement monitoring sub-module; among them:

[0071] The end deterioration feature visual recognition sub-module is used to identify the abnormal states of various components at the beam end, extract the apparent disease characteristics of the beam end components through non-contact visual analysis methods, and the disease indicators include at least scissor cross abnormality, beam end sleeper damage, beam end movable sleeper skew, beam end track spalling and chipping, beam end rail light band abnormality, beam end rail corrugation wear, and beam end fastener spring clip fracture; the apparent deterioration feature evaluation of the beam end area is carried out through the disease indicators;

[0072] The expansion device displacement recognition sub-module is used to monitor the representative displacement states of various components of the beam end expansion device, and the representative displacement states include at least the spacing of movable sleepers, the vertical displacement of steel sleepers, the rail expansion displacement, the expansion amount of longitudinal beams, and the scissor cross XZ-plane movement trajectory; the state of the expansion device is evaluated through the representative displacement states of the various components;

[0073] The beam end GNSS displacement monitoring sub-module is used to monitor the absolute longitudinal X displacement, absolute transverse Y displacement, and absolute vertical Z displacement of the beam end, and the beam end deck displacement is evaluated through the absolute longitudinal X displacement, absolute transverse Y displacement, and absolute vertical Z displacement of the beam end.

[0074] Specifically, the optoelectronic and high-definition camera integrated machine and GNSS Beidou in the deck inspection and monitoring device mainly complete the inspection and monitoring functions of the beam end deterioration feature visual recognition sub-module, the expansion device displacement recognition sub-module, and the beam end GNSS displacement monitoring sub-module. Now, further elaboration is as follows. First is the beam end deterioration feature visual recognition module, which is mainly realized by high-definition cameras carried by two brackets. The disease indicators mainly include scissor cross abnormality, beam end sleeper damage, beam end movable sleeper skew, beam end track spalling and chipping, beam end rail light band abnormality, beam end rail corrugation wear, beam end fastener spring clip fracture and other indicators. Among them, the end deterioration feature visual recognition sub-module is used to identify the abnormal states of various components at the beam end, extract the apparent disease characteristics of the beam end components through non-contact visual analysis methods. The specific methods include:

[0075] Obtain the disease feature data set of the railway bridge beam end, and preprocess the data set. The disease features include at least sleeper damage, light band abnormality, sleeper spalling and chipping, corrugation wear, and fish-scale corrugation;

[0076] Build and train a disease recognition neural network. The specific architecture schematic diagram is as Figure 5As shown, the recognition neural network model includes an input layer, a convolutional layer, a pooling layer, and a transposed convolutional layer. The input layer processes the initial input of the neural network and is used to receive the original image data from the monitoring system. The preprocessed data passed through the input layer will enter the subsequent convolutional layer as the basis for feature extraction; the convolutional layer processes the monitoring data through the sliding operation of the convolutional kernel, performs convolutional calculations on the premise of restricting the model complexity and reducing overfitting, extracts signal feature parameters, and the size of the intermediate matrix output after the convolutional calculation is as follows:

[0077]

[0078] where i is the input size collected by the monitoring system, p is the boundary value, s is the stride, and k is the convolutional kernel stride.

[0079] For example Figure 6 As shown, when the 2x2 convolutional kernel in the disease recognition neural network operates on the 3x3 input matrix, taking the 2x2 area in the upper left corner of the input matrix as the starting point, through the product operation and accumulation process between neurons, the first output value b11 is calculated; then the convolutional kernel moves to the next 2x2 area by a preset stride distance and repeats the same convolutional operation to obtain the output value b12; this operation process continues until the convolutional kernel matrix advances to the end of the input matrix, and finally a 2x2 output feature matrix is formed.

[0080] After the convolutional layer of the disease recognition neural network, max pooling processing is performed. After the pooling layer processing, the disease recognition neural network will perform convolution and pooling operations on the data again to further extract image features; after the convolutional calculation is completed, the data is imported into the transposed convolutional layer to map the small-resolution intermediate data to the large-resolution analysis image, and the deconvolution kernel is introduced into the analysis data to calculate and rearrange the data matrix, so as to restore to the original input data size; the calculation formula for the deconvolution size is as follows:

[0081] o' = s(i - 1) + k' - 2p'

[0082] where i is the size of the intermediate matrix of the monitoring system, p is the boundary value, s is the stride, and k is the deconvolution kernel stride;

[0083] The intermediate matrix that has completed the expansion of the deconvolution layer will be processed by the fully connected layer, express the prediction result in the form of a probability distribution, and complete the accurate determination of the beam-end disease category. The calculation formula of the fully connected layer is as follows:

[0084]

[0085] where O is the output matrix of the fully connected layer, I is the intermediate matrix input to the fully connected layer, w is the weight coefficient of the neurons in the corresponding layer, and B represents the bias of the fully connected layer;

[0086] After the fully connected processing is completed, the intermediate matrix is analyzed and normalized by the classifier, so that the unnormalized prediction of the intermediate matrix is transformed into non-negative numbers and the sum is 1. The formula for its normalized probability distribution is as follows:

[0087]

[0088] where n is the total number of neurons in this layer, o i represents the log probability of the i-th node;

[0089] Cross-entropy loss is used as the model evaluation quantization evaluation to express the fitting degree calculated by the recognition module. The formula of the cross-entropy evaluation function is as follows:

[0090]

[0091] where q(x i ) is the conclusion probability obtained through the disease recognition analysis and processing, and p(x i ) is the actual observed result.

[0092] The disease recognition neural network continuously iterates and updates the data set, uses the steepest descent method to optimize the parameters in the model, and analyzes the loss function and accuracy of the neural network model by real-time monitoring the performance of the neural network model to improve the model training accuracy; the boundary range of the model prediction and the actual boundary situation are analyzed as the investigation, and its precision, recall rate, and intersection over union are used as the evaluation indicators of the model accuracy performance; the evaluation formulas are as follows:

[0093]

[0094] where TP refers to the number of instances correctly identified as positive by the model, reflecting the correct judgment ability of the model for positive classes; TN refers to the number of instances correctly identified as negative by the model, reflecting the correct judgment ability of the model for negative classes; relatively, FP refers to the number of instances that the model wrongly judges negative instances as positive; FN refers to the number of instances that the model wrongly judges positive instances as negative;

[0095] The trained model is integrated into the edge computing gateway to process, extract features and identify damages for the beam end image information collected in real time in the main bridge and approach bridge areas, and conduct unified processing on the processed disease recognition information to form a disease assessment report for evaluating the severity of diseases in the beam end area, predicting the future severity and providing maintenance suggestions, and generating and storing a monitoring report for the recognition results.

[0096] In this embodiment, the specific working methods of the expansion device displacement recognition sub-module and the beam end GNSS displacement monitoring sub-module include:

[0097] Fix the beam surface inspection and monitoring device, and place the measuring point target on the switch rail, the left longitudinal beam of the rail lifting device, the left scissor cross, the stock rail, the movable sleeper, the right longitudinal beam of the rail lifting device, and the right scissor cross by surface mounting;

[0098] After the beam surface inspection and monitoring device is installed and fixed, carry out the device equipment debugging. Calibrate the initial values of the X, Y, and Z axes of the GNSS Beidou device in the GNSS displacement monitoring sub-module at the beam end. The change amounts in the X, Y, and Z directions are respectively x 17 、y 17 、z 17 ; The first optoelectronic deflection meter is responsible for the target monitoring areas of the left scissor cross, the stock rail, the movable sleeper, and the right scissor cross. The second optoelectronic deflection meter is responsible for the target monitoring areas of the switch rail, the left longitudinal beam of the rail lifting device, and the right longitudinal beam of the rail lifting device. The two optoelectronic deflection meters record the initial values of the X and Z axes of the targets in their respective responsible areas; Through the change amount x 17 in the X direction of the Beidou device, correct the actual longitudinal expansion and contraction displacement of the switch rail, the actual longitudinal expansion and contraction displacement of the left longitudinal beam of the rail lifting device, and the actual longitudinal expansion and contraction displacement of the right longitudinal beam of the rail lifting device respectively. Evaluate the target monitoring areas of the left scissor cross, the movable sleeper, and the right scissor cross by observing the change trend of the displacement value data in the X and Z directions.

[0099] Furthermore, elaborate on the expansion joint displacement identification sub-module and the GNSS displacement monitoring sub-module at the beam end. The expansion joint displacement identification sub-module mainly conducts real-time monitoring of relevant indicators through two optoelectronic deflection meters responsible for the main bridge and approach bridge areas. The optoelectronic deflection meter can monitor the planar displacement within the field of view. The main content indicators involved include the spacing of movable sleepers, the vertical displacement of steel sleepers, the rail expansion and contraction displacement (stock rail displacement and switch rail displacement), the longitudinal expansion amount of the longitudinal beam, and the XZ-plane movement trajectory of the scissor cross. The specific layout is as Figure 3 shown. The measuring point target is a reflector material target, which has the advantages of waterproof, high reflectivity, moisture-proof, durable and corrosion-resistant. It is placed on the 29-switch rail (longitudinal expansion and contraction displacement of the switch rail), 30-left longitudinal beam of the rail lifting device (longitudinal expansion amount of the longitudinal beam), 31 / 32 / 33-left scissor cross (XZ-plane movement trajectory), 34-stock rail (longitudinal expansion and contraction displacement of the stock rail), 35 / 36 / 37 / 38-movable sleeper (spacing of movable sleepers), 39-right longitudinal beam of the rail lifting device (longitudinal expansion amount of the longitudinal beam), 40 / 41 / 42-right scissor cross (XZ-plane movement trajectory) by surface mounting, avoiding the influence of drilling or welding installation equipment on the structure durability in the conventional monitoring objects.

[0100] After the beam surface inspection and monitoring device is installed and fixed, carry out the device equipment debugging. Calibrate the initial values of the X, Y, and Z axes of the GNSS Beidou device. The change amounts in the three directions are respectively x 17 、y17 , z 17 ; The 19 - photoelectric deflection meter (main bridge area) is mainly responsible for the target monitoring areas such as the 31 / 32 / 33 - left scissor lift, 34 - basic rail of the rail, 35 / 36 / 37 / 38 - movable sleepers, 40 / 41 / 42 - right scissor lift, etc. The 20 - photoelectric deflection meter (approach bridge area) is mainly responsible for the target monitoring areas such as the 29 - switch rail, 30 - left longitudinal beam of the rail lifting device, 39 - right longitudinal beam of the rail lifting device, etc. The two photoelectric deflection meters record the initial values of the X and Z axes of the targets in their respective responsible areas, and the original calibration values of each device are recorded in the embedded acquisition gateway database module.

[0101] Further, the calculation methods of each index on the beam surface are described.

[0102] The measured value of the longitudinal expansion of the 29 - switch rail is x' 29 , and the actual displacement change is X 29 :

[0103] X 29＝ x' 29 +x 17

[0104] The measured value of the longitudinal expansion of the 30 - left longitudinal beam of the rail lifting device is x' 30 , and the actual displacement change is X 30 :

[0105] X 30＝ x' 30 +x 17

[0106] The measured value of the longitudinal expansion of the 39 - right longitudinal beam of the rail lifting device is x' 39 , and the actual displacement change is X 39 :

[0107] X 39＝ x' 39 +x 17

[0108] The measured value of the longitudinal expansion of the 34 - basic rail of the rail is x' 34 , and the actual displacement change is X 34 :

[0109] X 34＝ x' 34 +x 17

[0110] For the target monitoring areas such as 31 / 32 / 33 - left scissor cross, 35 / 36 / 37 / 38 - movable sleeper, 40 / 41 / 42 - right scissor cross, etc., the main evaluation method is to observe whether the change trends of the data with each label are consistent, which is mainly reflected in the displacement values in the X and Z directions, and whether there are significant differences in the mean values of the data changes for each type of group. Taking the longitudinal X displacement of 31 / 32 / 33 - left scissor cross as an example to illustrate the method, the method for the vertical Z displacement is the same and will not be elaborated.

[0111] Within the time range T, the longitudinal displacement arrays corresponding to different node positions of the left scissor cross are X 31 , X 32 , X 33 , where the data set within X 31 is x 11 , x 12 , ··· x 1n , the data set within X 32 is x 21 , x 22 , ··· x 2n , the data set within X 33 is x 31 , x 32 , ··· x 3n , and the number of monitored values in each data group are n 1 , n 2 , n 3 .

[0112] Calculate the within - group variance and between - group variance of each data set. First, calculate the total mean value X 31 , X 32 , X 33 from the sample data of each group, and the specific calculation method is as follows: pt The specific calculation method is as follows:

[0113]

[0114] Calculate the between - group variance SSB. The between - group variance represents the sum of the squared deviations of the means of different groups from the total mean. Here, X pi is the mean of the i - th group, and the specific calculation method is as follows:

[0115]

[0116] Calculate the within - group variance SSE. The within - group variance represents the sum of the squared deviations of the observed values within each group from the mean of that group. Here, X pi is the mean of the i - th group, and the specific calculation method is as follows:

[0117]

[0118] Calculate the between-group mean square MSB and the within-group mean square MSE, where k is the number of groups, determined according to the set of measurement point quantities. Here, the value of k is taken as 3, and N is the total number of samples in the data set, N = n 1 + n 2 + n 3 , and the specific calculation method is as follows:

[0119]

[0120] Use the F statistic to test the ratio of the between-group variance to the within-group variance, look up the critical value corresponding to a significance level of 0.05 in the F distribution table, and compare it with the ratio of the between-group variance to the within-group variance. If the F value is large, it indicates that the between-group differences are large. Compared with the within-group differences, it shows that the means of each group of the longitudinal X displacement of the left scissor cross on 31 / 32 / 33 may have significant differences, and railway maintenance personnel need to further conduct on-site inspections to avoid abnormal components affecting train operation safety. If the F value is small, it indicates that the between-group differences are not much different from the within-group differences, and the longitudinal X displacement of the left scissor cross on 31 / 32 / 33 changes synergistically and the working state is normal.

[0121] The pier-beam state monitoring module is used to monitor the rotation angle and vibration of the pier top and the displacements of the main bridge and the approach bridge, and to evaluate each response index of the pier-beam;

[0122] In this embodiment, the pier-beam state monitoring module is used to monitor the rotation angle and vibration of the pier top and the displacements of the main bridge and the approach bridge. The specific method includes: by arranging displacement gauges, inclinometers, and vibration meters, evaluating the change states of the main bridge girder, the approach bridge girder, and the connecting pier components. The obtained indexes at least include the rotation angle of the pier top, the vibration of the pier top, the displacement of the main bridge girder, and the displacement of the approach bridge girder; the indexes involving displacement measurement in the test process are all integrated action quantities in each direction, and each single displacement component has a functional relationship with the displacement change of the pier top. The height of the pier top is h d , the longitudinal deflection angle of the pier top is θ x , θ x The clockwise deflection is positive, and the counterclockwise deflection is negative. When the actual pier top has a rotational displacement, both the vertical and longitudinal displacements of the pier top change.

[0123] A further description of the pier-beam state monitoring module is given. The pier-beam state monitoring module mainly evaluates the change states of components such as the main bridge girder, approach bridge girder, and connecting piers by arranging displacement gauges, inclinometers, and vibration gauges. The main content indicators include the pier top rotation angle of the pier (longitudinal rotation angle in the XZ plane and transverse rotation angle in the YZ plane), the pier top vibration of the pier (longitudinal vibration in the X-axis direction and transverse vibration in the Y direction), the displacement of the main bridge girder (longitudinal displacement of the beam joint in the X-axis, longitudinal displacement of the main bridge girder end support in the X-axis, and vertical displacement of the main bridge girder end in the Z-axis), and the displacement of the approach bridge girder (longitudinal displacement of the approach bridge girder end support in the X-axis and vertical displacement of the approach bridge girder end in the Z-axis). Each index is used to evaluate the response indexes of the pier and beam. The schematic diagram of the pier-beam state monitoring module is as shown in Figure 7 shown. The descriptions of each label are as follows: 6-1 left approach bridge girder displacement gauge, 6-2 right approach bridge girder displacement gauge, 9-1 pier top rotation angle of the pier, 9-2 pier top vibration of the pier, 10-1 left main bridge girder displacement gauge, 10-2 right main bridge girder displacement gauge, 11-1 left beam joint displacement gauge, 11-2 right beam joint displacement gauge.

[0124] Since the indexes involved in the displacement measurement during the testing of each index are the integrated action quantities in each direction, rather than the single variables of each component, each single displacement component has a functional relationship with the change of the pier top displacement. The height of the pier top is h d , the longitudinal deflection angle of the pier top is θ x , θ x is positive for clockwise deflection and negative for counterclockwise deflection. When the actual pier top has a rotational displacement, both the vertical and longitudinal displacements of the pier top change. Considering that when the actual pier is displaced, the deflection angle is small and the pier stiffness is large, each direction can be approximately rigidly changed. The functional relationship is as shown in Figure 8 shown.

[0125] To further accurately and real-time analyze the corresponding components of each index, the calculation methods of each index of the pier-beam state are further described.

[0126] When the pier is displaced, the vertical displacement of the pier top is △z, and the longitudinal displacement of the pier top is △x:

[0127] △z = h d -|h d × cosθ x |

[0128] △x = h d × sinθ x

[0129] The longitudinally measured value of the left approach bridge girder displacement gauge 6-1 is x’ 6-1 , and the actual changed displacement is X 6-1 :

[0130] X 6-1 = x’ 6-1 -h d×sinθ x

[0131] The longitudinally measured value of the main girder displacement of the approach bridge on the right side of 6-2 is x' 6-2 , and the actual displacement change is X 6-2 :

[0132] X 6-2 = x' 6-2 - h d ×sinθ x

[0133] The longitudinally measured value of the main girder displacement of the main bridge on the left side of 10-1 is x' 10-1 , and the actual displacement change is X 10-1 :

[0134] X 10-1 = x' 10-1 - h d ×sinθ x

[0135] The vertically measured value of the main girder displacement of the main bridge on the left side of 10-1 is z' 10-1 , and the actual displacement change is Z 10-1 :

[0136] Z 10-1 = z' 10-1 -(h d -|h d ×cosθ x |)

[0137] The longitudinally measured value of the main girder displacement of the main bridge on the right side of 10-2 is x' 10-2 , and the actual displacement change is X 10-2 :

[0138] X 10-2 = x' 10-2 - h d ×sinθ x

[0139] The vertically measured value of the main girder displacement of the main bridge on the right side of 10-2 is z' 10-2 , and the actual displacement change is Z 10-2 :

[0140] Z 10-2 = z' 10-2 -(h d -|h d ×cosθ x |)

[0141] The 11-1 left beam joint displacement gauge and the 11-2 right beam joint displacement gauge mainly measure the beam joint displacement. The sensors are mainly fixed at the ends of the approach bridge and the main bridge. The beam joint displacement change is not affected by the pier top deviation.

[0142] The network transmission and power supply module is used for data transmission and on-line power supply to the beam-end environment monitoring module, the beam surface state monitoring module and the pier-beam state monitoring module, ensuring smooth data and image transmission and stable voltage.

[0143] Specifically, the network transmission and power supply module includes an integrated processing sub-module, a private network transmission sub-module and a voltage stabilization power supply sub-module; the integrated processing sub-module synchronously collects the data of each module by means of line ring network integration, and uses an industrial switch for internal ring network communication transmission. When a single device fails, the signal data can continue to be transmitted along the ring link, realizing the function that the damage of a single component does not affect the overall operation of the device; the private network transmission sub-module and the voltage stabilization power supply sub-module transmit and supply power to the test data of each module through an edge computing gateway.

[0144] This embodiment discloses a state monitoring system for the beam-end area of a long-span railway bridge, including: a beam-end environment monitoring module, a beam surface state monitoring module, a pier-beam state monitoring module, and a network transmission and power supply module; the beam-end environment monitoring module is used for monitoring the environmental characteristics of the beam-end of a long-span railway bridge, and the environmental characteristics specifically include the wind speed, wind direction, environmental temperature and humidity of the beam-end bridge deck; the beam surface state monitoring module is used for monitoring various index states of the beam-end beam surface, realizing the evaluation of the apparent deterioration characteristics of the beam-end area, the evaluation of the state of the beam-end expansion device, and the evaluation of the displacement of the beam-end beam surface; the pier-beam state monitoring module is used for monitoring the pier top rotation and vibration of the pier and the displacement of the main bridge and the approach bridge, realizing the evaluation of each response index of the pier-beam; the network transmission and power supply module is used for data transmission and on-line power supply to the beam-end environment monitoring module, the beam surface state monitoring module and the pier-beam state monitoring module, ensuring smooth data and image transmission and stable voltage.

[0145] The state monitoring system for the beam-end area of a long-span railway bridge disclosed in this embodiment is applied to the beam-ends of different bridge structures such as continuous beam bridges, arch bridges, cable-stayed bridges, and suspension bridges of long-span railway bridges, and can significantly improve the accuracy, efficiency and safety of the state health management of the beam-ends of bridges. Through functions such as real-time monitoring, disease identification, data analysis, and fault warning, it can effectively prevent the deterioration of bridge diseases, extend the service life of bridges, reduce maintenance costs, and ensure the smooth operation of railway traffic. With the continuous development of intelligent technologies, the application prospects of the monitoring system and method of the present invention are broad, and it will play an important role in the field of intelligent transportation in the future, promoting the management of railway bridges towards a more scientific and refined management mode.

[0146] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of the present disclosure. The appended method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy recited.

[0147] In the above detailed description, various features are combined in a single embodiment to simplify the present disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are expressly recited in each claim. Rather, as reflected in the appended claims, the invention lies in less than the full scope of the features of the individual disclosed embodiments. Accordingly, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0148] Those skilled in the art should also understand that the various illustrative logical blocks, modules, circuits, and algorithmic steps described in connection with the embodiments herein can be implemented as electronic hardware, computer software, or combinations thereof. To clearly illustrate the interchangeability of hardware and software, the various illustrative components, blocks, modules, circuits, and steps have been generally described in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in a variable manner for each particular application, but such implementation decisions should not be interpreted as departing from the scope of the present disclosure.

[0149] The steps of a method or algorithm described in connection with the embodiments herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination thereof. The software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. Of course, the storage medium can also be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. Of course, the processor and the storage medium can also exist as discrete components in a user terminal.

[0150] For software implementation, the techniques described in this application can be implemented by modules (e.g., procedures, functions, etc.) that perform the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented within the processor or outside the processor. In the latter case, it is communicatively coupled to the processor via various means, which are well known in the art.

[0151] The above description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but those of ordinary skill in the art should recognize that the various embodiments can be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. In addition, with respect to the term "comprising" as used in the specification or claims, this term is covered in a manner similar to the term "including," as interpreted when "including" is used as a transitional word in a claim. Further, any use of the term "or" in the specification or claims is to mean "non-exclusive or."

Claims

1. A long-span railway bridge beam end area status inspection and monitoring system, characterized in that: include: Beam end environment monitoring module, beam surface status monitoring module, pier beam status monitoring module, network transmission and power supply module; The beam end environment monitoring module is used to monitor the environmental characteristics of the beam end of a long-span railway bridge, and the environmental characteristics specifically include wind speed, wind direction, ambient temperature and humidity on the bridge deck at the beam end; The beam surface status inspection and monitoring module is used to inspect and monitor the status of various indicators of the beam end and beam surface, to evaluate the apparent degradation characteristics of the beam end area, to evaluate the status of the beam end expansion device, and to evaluate the displacement of the beam end and beam surface; The pier and beam status monitoring module is used to monitor the rotation angle and vibration of the pier top, the displacement of the main bridge and the approach bridge, and to evaluate the response indicators of the piers and beams; The network transmission and power supply module is used for data transmission and online power supply of the beam end environment monitoring module, beam surface status monitoring module and pier beam status monitoring module, ensuring smooth data and image transmission and stable voltage.

2. A long-span railway bridge beam end area status inspection and monitoring system as claimed in claim 1, characterized in that: The beam end environment monitoring module includes a wind speed and direction sensor and a temperature and humidity sensor. The wind speed and direction sensor is installed on the cross arm flange perpendicular to the longitudinal direction of the bridge deck to test the wind speed and direction environment in the beam end area. The temperature and humidity sensor is installed at the connection between the horizontal arm and the vertical pole to test the temperature and humidity environment conditions in the beam end area.

3. A long-span railway bridge beam end area status inspection and monitoring system as claimed in claim 1, characterized in that: The beam surface status inspection and monitoring module includes an end degradation feature visual recognition submodule, a telescopic device displacement recognition submodule, and a beam end GNSS displacement monitoring submodule; wherein: The end degradation feature visual recognition submodule is used to identify the abnormal state of each component at the beam end, and extract the apparent disease characteristics of the beam end components through a non-contact visual analysis method. The disease indicators include at least abnormal scissor fork, damaged rail sleeper at the beam end, skewed movable rail sleeper at the beam end, peeling and falling of rail at the beam end, abnormal light band of the rail at the beam end, wave wear of the rail at the beam end, and broken spring bars of the beam end fasteners; the disease indicators are used to evaluate the apparent degradation characteristics of the beam end area; The telescopic device displacement identification submodule is used to monitor the representative displacement states of each component of the beam end telescopic device, wherein the representative displacement states at least include the movable sleeper spacing, the vertical displacement of the sleeper, the telescopic displacement of the rail, the telescopic amount of the longitudinal beam, and the XZ plane activity trajectory of the scissors fork; the telescopic device state is evaluated through the representative displacement states of each component; The beam end GNSS displacement monitoring submodule is used to monitor the longitudinal X absolute displacement of the beam end, the transverse Y absolute displacement of the beam end and the vertical Z absolute displacement of the beam end, and evaluate the beam end beam surface displacement through the longitudinal X absolute displacement of the beam end, the transverse Y absolute displacement of the beam end and the vertical Z absolute displacement of the beam end.

4. A long-span railway bridge beam end area status inspection and monitoring system as claimed in claim 3, characterized in that: The end degradation feature visual recognition submodule is used to identify the abnormal state of each component at the beam end and extract the apparent disease characteristics of the beam end components through non-contact visual analysis methods. The specific methods include: Obtaining a data set of railway bridge beam end disease characteristics, and preprocessing the data set, wherein the disease characteristics at least include sleeper damage, light band abnormality, sleeper peeling, wave wear, and fish scale ripples; Build and train a disease recognition neural network. The recognition neural network model includes an input layer, a convolution layer, a pooling layer, and a transposed convolution layer. The input layer processes the initial input of the neural network and is used to receive the original image data from the monitoring system. The pre-processed data of the input layer will enter the subsequent convolution layer as the basis for feature extraction; the convolution layer processes the monitoring data through the sliding operation of the convolution kernel, performs convolution calculations under the premise of limiting the complexity of the model and reducing overfitting, and extracts signal feature parameters. The size of the intermediate matrix output after the convolution calculation is as follows: Where i is the input size collected by the monitoring system, p is the boundary value, s is the step size, and k is the convolution kernel step size.

5. A long-span railway bridge beam end area status inspection and monitoring system as claimed in claim 4, characterized in that: When the 2x2 convolution kernel in the disease recognition neural network operates in an input matrix of size 3x3, the 2x2 area in the upper left corner of the input matrix is ​​used as the starting point, and the first output value b11 is calculated through the multiplication and accumulation process between neurons; then the convolution kernel moves to the right by a preset step length to reach the next 2x2 area, and repeats the same convolution operation to obtain the output value b12; the operation process continues until the convolution kernel matrix advances to the end of the input matrix, and finally forms a 2x2 output feature matrix.

6. A long-span railway bridge beam end area status inspection and monitoring system as claimed in claim 4, characterized in that: The convolution layer of the disease recognition neural network is followed by the maximum pooling process. After the pooling layer, the disease recognition neural network will perform convolution and pooling operations on the data again to further extract image features. After the convolution calculation is completed, the data is imported into the transposed convolution layer, the low-resolution intermediate data is mapped to the high-resolution analysis image, and the deconvolution kernel is introduced into the analysis data for calculation and rearrangement of the data matrix, thereby restoring the original input data size. The calculation formula of the deconvolution size is as follows: o'=s(i-1)+k'-2p' Where i is the size of the intermediate matrix of the monitoring system, p is the boundary value, s is the step size, and k is the step size of the deconvolution kernel; The intermediate matrix after the deconvolution layer expansion will be processed by the fully connected layer, and the prediction results will be expressed in the form of probability distribution to accurately determine the type of beam end defects. The calculation formula of the fully connected layer is as follows: Among them, O is the output matrix of the fully connected layer, I is the intermediate matrix of the input fully connected layer, w is the weight coefficient of the corresponding layer neuron, and B represents the bias of the fully connected layer; After the full connection processing is completed, the intermediate matrix is ​​analyzed and normalized by the classifier, so that the unnormalized prediction of the intermediate matrix is ​​transformed into a non-negative number and the sum is 1. The normalized probability distribution calculation formula is as follows: Where n is the total number of neurons in this layer, o i represents the logarithmic probability of the i-th node; The cross entropy loss is used as a quantitative evaluation of the model to express the degree of fit calculated by the recognition module. The formula of the cross entropy judgment function is as follows: where q(x i ) is the conclusion probability obtained through disease identification and analysis, p(x i ) is the result of actual observation.

7. A long-span railway bridge beam end area condition detection monitoring system as claimed in claim 6, characterized in that: The disease recognition neural network continuously iterates and updates the data set, optimizes the parameters in the model using the steepest descent method, and improves the model training accuracy by real-time monitoring of the neural network model performance, analyzing the neural network model loss function and the neural network model accuracy; The analysis finally uses the boundary range of the model prediction and the boundary of the actual situation as the examination, and examines its precision, recall rate and intersection-over-union ratio as the evaluation indicators of the model accuracy performance; the evaluation formulas are as follows: Among them, TP refers to the number of instances correctly identified as positive by the model, reflecting the model's ability to correctly judge the positive class; TN refers to the number of instances correctly identified as negative by the model, reflecting the model's ability to correctly judge the negative class; relatively, FP refers to the number of negative class instances that the model mistakenly judges as positive; FN refers to the number of positive class instances that the model mistakenly judges as negative; The trained model is integrated into the edge computing gateway to process, extract features and identify damage of beam end images collected in real time in the main bridge and approach bridge areas. The processed disease identification information is integrated to form a disease assessment report, which is used to assess the severity of diseases in the beam end area, predict the future severity and provide maintenance recommendations, and generate and store the identification results into a monitoring report.

8. A long-span railway bridge beam end area condition inspection and monitoring system as claimed in claim 3, characterized in that: The specific working methods of the telescopic device displacement identification submodule and the beam end GNSS displacement monitoring submodule include: Fix the beam surface inspection monitoring device, and place the measuring point targets on the rail point rail, the left longitudinal beam of the rail lifting device, the left scissor fork, the rail base rail, the movable sleeper, the right longitudinal beam of the rail lifting device and the right scissor fork by surface sticking; After the beam surface inspection monitoring device is installed and fixed, the device is debugged and the initial values ​​of the X, Y, and Z axes of the GNSS Beidou equipment in the beam end GNSS displacement monitoring submodule are calibrated. The changes in the X, Y, and Z directions are x and y, respectively. 17 ,y 17 、z 17 The first photoelectric deflectometer is responsible for the left scissor fork, rail base rail, movable rail sleeper, and right scissor fork target monitoring area. The second photoelectric deflectometer is responsible for the rail point rail, left longitudinal beam of the rail lifting device, and right longitudinal beam of the rail lifting device. The two photoelectric deflectometers record the initial values ​​of the X and Z axes of the targets in each responsible area. The change in the X direction of the Beidou device x 17 The actual changes in the longitudinal expansion and contraction of the rail point rail, the actual changes in the longitudinal expansion and contraction of the left longitudinal beam of the rail lifting device, and the actual changes in the longitudinal expansion and contraction of the right longitudinal beam of the rail lifting device are corrected respectively. By observing the change trends of the displacement value data in the X and Z directions, the target monitoring areas of the left scissors fork, movable sleeper and right scissors fork are evaluated.

9. A long-span railway bridge beam end area condition detection monitoring system as claimed in claim 1, characterized in that: The pier-beam status monitoring module is used to monitor the rotation angle and vibration of the pier top, and the displacement of the main bridge and the approach bridge. The specific method includes: by arranging displacement meters, inclinometers, and vibrometers, the main bridge beam, the approach bridge beam, and the connection pier components are evaluated. The indicators obtained include at least the rotation angle of the pier top, the vibration of the pier top, the displacement of the main bridge beam, and the displacement of the approach bridge beam; the indicators involved in the displacement measurement during the test are all integrated actions in all directions, and each single displacement component has a functional relationship with the displacement change of the pier top. The pier top height is h. d , the pier top deflects longitudinally at an angle of θ x ,θ x Clockwise deflection is positive, counterclockwise deflection is negative. When the actual pier top rotates and deflects, both the vertical and longitudinal displacements of the pier top change.

10. A long-span railway bridge beam end area condition inspection and monitoring system as claimed in claim 1, characterized in that: The network transmission and power supply module includes an integrated processing submodule, a private network transmission submodule and a voltage-stabilizing power supply submodule; the integrated processing submodule adopts an integrated line ring network to synchronously collect data from each module, and uses an industrial switch for internal ring network communication transmission. When a single device fails, the signal data can continue to be transmitted along the ring link, thereby realizing the function that the damage of a single component does not affect the overall operation of the equipment; the private network transmission submodule and the voltage-stabilizing power supply submodule transmit the test data of each module and provide power supply support through the edge computing gateway.

Images