Dynamic vision-based bridge overall displacement monitoring and behavior twin analysis method and system

Abstract

The present invention belongs to the technical field of bridge monitoring, and specifically discloses a dynamic visual bridge overall displacement monitoring and behavior twin analysis method and system. It includes obtaining bridge component displacement monitoring data; performing time domain feature and frequency domain feature analysis on the displacement monitoring data; establishing a displacement response model under different working conditions based on the inherent change law of bridge structure displacement under different external conditions to identify abnormal displacement patterns; calculating the spatial correlation between different components to analyze the interaction and influence of different components under dynamic loads; predicting the change trend of bridge displacement data based on the ARIMA time series model; using causal inference methods to analyze the reasons for the displacement changes of different components and identify key influencing factors. The present invention realizes the fusion and comprehensive utilization of displacement monitoring data by analyzing the spatial distribution characteristics and time evolution characteristics of the displacement data of key components, which is of great significance to the health monitoring of bridge structures.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge monitoring, and more specifically, relates to a dynamic visual bridge overall displacement monitoring and behavior twin analysis method and system. Background Art

[0002] As the service time of bridges increases, under the long-term action of external loads, many bridges will experience various structural damages and continue to accumulate, and their bearing capacity and safety will decrease, which may cause serious structural safety accidents. In order to ensure the safety and durability of bridges in service, bridge management personnel usually conduct regular inspections or long-term monitoring of bridge structures, and use the bridge information obtained as the basis for management and maintenance decisions, providing a reference for operation and maintenance.

[0003] In bridge structural health monitoring, monitoring items can generally be divided into two categories: load (vehicle load, ambient temperature and humidity, wind load, etc.) and response (vibration acceleration, displacement, strain, etc.), which reflect the size of the load borne by the bridge and the response of the local or global structure of the bridge to the load. Among the structural response indicators, stress and strain reflect the local response of the structure. The stress and strain changes caused by external forces are often concentrated near the external force application point. Therefore, in the case of sparsely arranged strain or stress sensors, stress and strain monitoring is mainly used to assess whether key components are damaged rather than the overall health of the bridge. Displacement, as a key indicator for assessing whether a structure is safe, reflects the overall deformation of the structure. It is a comprehensive reflection of indicators such as the stiffness and structural bearing capacity of the bridge, and represents the overall health of the bridge. If its size exceeds a certain limit, it means that the bridge condition may be abnormal or risky. Therefore, accurate measurement of displacement is crucial for efficient structural health monitoring.

[0004] The displacement measurement of bridges is mostly based on draw-rope displacement sensors, static levels, and GNSS methods. Draw-rope displacement sensors measure the displacement of the draw rope to indirectly measure the displacement of the structure. They are more suitable for measuring the opening and closing displacement between two components, but not suitable for measuring the vertical displacement of the bridge main beam. Static levels indirectly measure the displacement of the structure by measuring the height change of the liquid. They have high measurement accuracy but can only measure vertical displacement. The liquid surface in the tube reacts laggingly to the dynamic displacement and has a low sampling frequency, making them more suitable for measuring static displacement. The GNSS method is based on displacement measurement based on the global positioning system, but this method requires that there should be no obstructions near the equipment layout, and the vertical displacement measurement accuracy is relatively low and is susceptible to electromagnetic interference.

[0005] In addition, the above methods require the installation of sensors and corresponding circuits and other supporting facilities at each measuring point on the bridge to carry out large-scale real-time monitoring, which places high demands on engineering budget and construction. Traditional displacement monitoring methods usually rely on a limited number of sensors, which may result in monitoring results that only reflect the local displacement of the structure and fail to provide a comprehensive understanding of the overall structural behavior. Sensor-based displacement monitoring is usually focused on specific monitoring points, which may lead to the omission of some potential problem areas. The deformation of the overall structure may not be fully evaluated through limited monitoring points.

[0006] In view of the problems existing in traditional displacement monitoring, this project proposes a twin analysis method of bridge overall displacement monitoring and displacement behavior based on dynamic vision. Dynamic visual monitoring completes comprehensive monitoring of the displacement of the entire bridge structure, rather than being limited to specific monitoring points. The arrangement of cameras can cover different areas of the bridge and capture the displacement of the overall structure. Compared with some traditional displacement monitoring methods, the monitoring method based on dynamic vision does not require adding a large number of sensors or equipment to the structure, and can choose to set up measurement stations in areas where it is more suitable to lay cables. Therefore, it has flexible layout, greatly reduced installation costs, and less interference to the structure. The monitoring system based on dynamic vision can achieve high-sampling frequency image acquisition, thereby completing high-sampling frequency displacement monitoring. At the same time, using computer vision technology, more advanced and more visual data processing and analysis can be performed, including target detection, image processing, feature point tracking, etc., so as to extract information on the status of multiple related bridge components. Further, the displacement behavior of bridge structures such as main beams, supports, expansion joints, dampers, and suspenders under different working conditions is analyzed intensively, and a twin model of the overall displacement behavior of the whole bridge is constructed to deepen the understanding of the overall displacement behavior of the bridge.

[0007] Therefore, taking bridges as the measurement objects, this field urgently needs to propose a method for monitoring the overall displacement of bridges and twin analysis of displacement behavior based on dynamic vision, so as to achieve high-precision, high sampling frequency, low-cost, automated measurement of structural displacement that meets the requirements of long-term online monitoring, and perform twin analysis of displacement behavior, which has important practical significance for structural health monitoring. Summary of the invention

[0008] In response to the above defects or improvement needs of the prior art, the present invention provides a dynamic visual bridge overall displacement monitoring and behavioral twin analysis method, which comprehensively uses statistical analysis, time series analysis, machine learning and other methods to study the spatial distribution characteristics and temporal evolution characteristics of the displacement data of key components, realize the fusion and comprehensive utilization of displacement monitoring data, and provide a data information basis for subsequent more comprehensive twin status assessment.

[0009] To achieve the above object, according to one aspect of the present invention, a dynamic visual bridge overall displacement monitoring and behavior twin analysis method is proposed, comprising the following steps:

[0010] Step 1, obtaining bridge component displacement monitoring data;

[0011] Step 2: Analyze the time domain characteristics and frequency domain characteristics of the displacement monitoring data;

[0012] Step 3: Based on the inherent variation law of bridge structure displacement under different external conditions, a displacement response model under different working conditions is established to identify abnormal displacement patterns;

[0013] Step 4: Calculate the spatial correlation between different components to analyze the interaction and influence of different components under dynamic loads;

[0014] Step 5: Use causal inference methods to analyze the causes of displacement changes of different components and identify key influencing factors;

[0015] Step six: Based on the ARIMA time series model, predict the changing trend of bridge displacement data.

[0016] As further preferred, in step 3, identifying the abnormal displacement pattern includes:

[0017] Based on the Z-score statistical method, the time domain and frequency domain change trends of the monitored displacement data are analyzed to identify abnormal displacement patterns.

[0018] Preferably, step three further includes: constructing a variation trend model of the displacement time history data in the time domain and the frequency domain:

[0019]

[0020] in, and They are i The mean and standard deviation of the displacement of each component are , then it is considered that the component is at time t An exception occurs. θ is the preset threshold, otherwise, the component t No abnormalities were observed.

[0021] As further preferred, in step 4, calculating the spatial correlation between different components includes:

[0022]

[0023] in, and They are i and jComponents at time t The displacement of and They are i and j The average displacement of the components, T is the total number of time steps, is the spatial correlation between the i-th component and the j-th component.

[0024] Preferably, in step 5, the change trend of the predicted displacement data includes:

[0025]

[0026] in, is a point in time t The predicted displacement value of is the intercept, is the autoregressive coefficient, is the displacement monitoring value of the point at time t-1, t-2, ..., tp or the maximum and average displacement values ​​in the nearby interval, is the error term.

[0027] As a further preferred embodiment, in step five, a causal inference method is used to analyze the causes of displacement changes of different components, identify key influencing factors, and based on the Granger causality test, determine whether a component displacement time series is helpful in predicting another time series.

[0028] As a further preferred embodiment, judging whether a component displacement time series is helpful to predict another time series based on Granger causality test includes:

[0029] Step S51: Check the displacement time series of the i-th component and the jth component displacement time series ;

[0030] Step S52, set the original hypothesis : displacement of the ith component Not the jth component displacement Granger reasons;

[0031] Step S53, constructing a simplified model of displacement prediction value:

[0032]

[0033] in, is the jth component at time t The predicted displacement value of is the intercept, is the autoregressive coefficient, is the displacement monitoring value of the jth component at time t-1, t-2, …, tp, is the error term;

[0034] Step S54, constructing a complete model of displacement prediction value:

[0035]

[0036] in, is the jth component at time t The predicted displacement value of is the intercept, is the regression coefficient, is the displacement monitoring value of the jth component at time t-1, t-2, …, tp, is the displacement monitoring value of the i-th component at time t-1, t-2, …, tp, is the error term;

[0037] Step S55, calculate the F statistic to test the null hypothesis:

[0038]

[0039] in, and are the residual sums of squares of the reduced model and the full model, p is the number of additional lags, T is the sample size, k is the number of parameters of the full model;

[0040] Step S56, according to the distribution of the F statistic, use the F distribution table to calculate and look up the corresponding probability value, and judge whether one component displacement time series is helpful to predict another time series according to the probability value.

[0041] Step S56 also includes:

[0042] If the probability value x is less than 0.05, the original hypothesis is rejected and the displacement of the i-th component is considered is the displacement of the jth component Granger reasons;

[0043] If the probability value x ≥ 0.05, the null hypothesis cannot be rejected and the displacement of the i-th component is considered Not the jth component displacement Granger reasons.

[0044] According to another aspect of the present invention, a dynamic visual bridge overall displacement monitoring and behavior twin analysis system is also provided, comprising:

[0045] The first main control module is used to obtain the displacement monitoring data of the bridge components;

[0046] A second main control module is used to analyze the time domain characteristics and frequency domain characteristics of the displacement monitoring data;

[0047] The third main control module is used to establish a displacement response model under different working conditions based on the inherent variation law of the bridge structure displacement under different external conditions to identify abnormal displacement patterns;

[0048] The fourth main control module is used to calculate the spatial correlation between different components to analyze the interaction and influence of different components under dynamic loads;

[0049] The fifth main control module is used to predict the changing trend of bridge displacement data based on the ARIMA time series model;

[0050] The sixth main control module is used to analyze the causes of displacement changes of different components and identify key influencing factors using causal inference methods.

[0051] As further preferred, the first main control module is arranged on the bridge, comprising:

[0052] A plurality of cameras disposed on the bridge for acquiring images of bridge components;

[0053] An image acquisition unit, used for receiving images acquired by a camera;

[0054] The image and data processing unit is used to process images to obtain bridge component displacement monitoring data.

[0055] As further preferred, the first main control module further includes:

[0056] The camera orientation calibration platform is used to calibrate the camera orientation so that a complete component image can be obtained by cooperating with multiple cameras;

[0057] Image synchronization acquisition module, used to synchronize the working time of all cameras.

[0058] In general, the above technical solution conceived by the present invention has the following technical advantages compared with the prior art:

[0059] 1. This invention proposes a method for monitoring the overall displacement of a bridge and twin analysis of its displacement behavior based on dynamic vision, which can achieve high-precision, high sampling frequency, low-cost, automated measurement of structural displacement that meets the requirements of long-term online monitoring, and perform twin analysis of displacement behavior, which has important practical significance for structural health monitoring.

[0060] 2. The present invention can realize high-precision and high-frequency monitoring of the overall and local dynamic displacement of the bridge, with a sampling frequency higher than 10Hz and a displacement pixel accuracy lower than 0.01pixel. The actual accuracy is related to the distance between the device and the bridge. A single device can complete the non-contact measurement of the overall displacement of the bridge, saving equipment and installation costs. The device can collect and process data in real time and stably. It is deployed outside the bridge area, easy to debug and maintain, and supports remote access, which is convenient for maintenance unit management.

[0061] 3. Based on the overall displacement data of the bridge at the same time, the present invention can determine the displacement behavior of bridge structures such as main beams, supports, expansion joints, dampers, and cables under different working conditions, which is conducive to studying the correlation and spatial structural relationship between the displacement responses of main cables, cables, main beams, and main towers in time / space and static and dynamic properties, and more comprehensively understand the overall performance and behavior patterns of the bridge. Constructing a twin model of the overall displacement behavior of the whole bridge deepens the understanding of the overall displacement behavior of the bridge and provides a scientific basis for preventive maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a dynamic visual bridge overall displacement monitoring and displacement behavior twin analysis method of the present invention;

[0063] Figure 2 It is a schematic diagram of a twin analysis method involved in the present invention;

[0064] Figure 3 It is a structural schematic diagram of a dynamic visual bridge overall displacement monitoring and displacement behavior twin analysis system of the present invention.

[0065] In all the drawings, the same reference numerals represent the same technical features, specifically: 1-suspension bridge; 2-image acquisition range of a single camera; 3-camera; 4-camera orientation calibration table; 5-image synchronization acquisition unit; 6-image and data processing unit; 7-wireless transmission unit; 8-housing; 9-bracket; 10-cable. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0067] like Figure 3As shown, an embodiment of the present invention provides a method for monitoring the overall displacement of a bridge and analyzing the displacement behavior of a twin based on dynamic vision, which is mainly realized by a visual monitoring station device, which is arranged in the vicinity of the bridge to ensure that there is no obstruction within the shooting range. The visual monitoring station device is composed of multiple cameras, multiple camera orientation calibration platforms, image synchronization acquisition modules, image and data processing modules, wireless transmission modules, housings, brackets, cables, etc.

[0068] Taking the monitoring scenario of a suspension bridge as an example, the visual monitoring station can be arranged in the area around the bridge, and the images of key positions such as the support, suspenders, main cables, and main beam quarter points / half points can be collected through multiple built-in cameras. The closer the monitoring station is to the bridge, the larger the image acquisition range of a single camera. When the distance is close enough and the appropriate lens is selected, the image acquisition of multiple key components can be achieved through a single camera. Based on image and data processing algorithms, the visual monitoring station compares the image observation points of adjacent frames of the same camera to extract relative pixel displacement, and converts it into bridge displacement based on the camera calibration results.

[0069] The camera is responsible for collecting images of the bridge displacement monitoring points and is fixed to the camera orientation calibration platform bracket by bolt compression; the camera orientation calibration platform is equipped with an angle adjustment device, which is responsible for adjusting the camera's shooting area and is fixed to the partition plate in the metal shell by bolts to prevent slippage due to external disturbances and ensure that the camera's shooting area is relatively fixed; the image synchronization acquisition module is connected to the camera through trigger lines and transmission lines, and realizes external triggering of image acquisition of multiple cameras through high and low levels, avoiding image frame loss in a physically-determined manner, and accurately synchronizing the time of image acquisition by all cameras to ensure the simultaneity and correlation of displacement data; the image and data processing module is responsible for storing and processing the images collected by each camera, Analysis and processing, based on the various algorithms installed, calculate the displacement changes of key points between adjacent frames of the same measuring point, obtain the displacement values ​​of each component at the same time, realize the overall displacement extraction of the bridge based on dynamic vision, and record the displacement time-series data in a standard format; the wireless transmission module is responsible for regularly transmitting the displacement data rather than the image, which is more economical and has an integrated remote access function so that users can remotely monitor the system status and data; the outer shell is responsible for protecting the camera and other hardware, preventing the internal hardware from being affected by environmental factors such as bad weather, dust, and humidity, to ensure its long-term stable operation; the dynamic vision monitoring station equipment is powered by external cables; the heat dissipation module, temperature and humidity sensors, etc. are not listed in detail.

[0070] Furthermore, a dynamic visual bridge overall displacement monitoring and behavior twin analysis system includes: a first main control module, used to obtain bridge component displacement monitoring data; a second main control module, used to perform time domain feature and frequency domain feature analysis on the displacement monitoring data; a third main control module, used to establish a displacement response model under different working conditions based on the inherent change law of bridge structure displacement under different external conditions, so as to identify abnormal displacement patterns; a fourth main control module, used to calculate the spatial correlation between different components, so as to analyze the interaction and influence of different components under dynamic loads; a fifth main control module, used to predict the change trend of bridge displacement data based on the ARIMA time series model; a sixth main control module, used to use causal inference methods to analyze the causes of displacement changes of different components and identify key influencing factors.

[0071] Optionally, the first main control module is arranged on the bridge, and includes: multiple cameras arranged on the bridge, used to obtain images of bridge components; an image acquisition unit, used to receive pictures obtained by the camera; and an image and data processing unit, used to process the pictures to obtain bridge component displacement monitoring data.

[0072] Optionally, the first main control module also includes: a camera orientation calibration platform for calibrating the camera orientation so that a complete component image can be acquired with the cooperation of multiple cameras; and an image synchronization acquisition module for synchronizing the working time of all cameras.

[0073] like Figure 1 and Figure 2 As shown in the figure, the implementation process of the bridge overall displacement monitoring and displacement behavior twin analysis method based on dynamic vision is as follows: measurement point selection, hardware selection, equipment layout, image acquisition, image preprocessing, displacement data extraction, displacement data feedback, and bridge overall displacement behavior twin analysis. Specifically:

[0074] Selection of measuring points: Fully study the overall structural characteristics of the bridge, ensure that the monitoring objects cover a variety of bridge structures, and determine the bridge measuring points according to the monitoring needs. Determine the location of the visual monitoring station through on-site surveys and drawings, ensure that there are no obstructions near the equipment layout point, the shooting distance and angle are appropriate, and the equipment can remain stable during displacement monitoring, so that higher quality images can be obtained.

[0075] Hardware selection: Determine the camera and lens selection based on the distance between the deployment location and the bridge monitoring target point. Determine the hardware selection of the image synchronization acquisition module, image and data processing module according to the algorithm requirements, sampling frequency and number of cameras, complete the visual monitoring station equipment integration, and ensure the real-time processing of images and the extraction of the overall displacement data of the bridge. Reasonable design of shell equipment installation: Install the bracket at the predetermined visual monitoring station deployment location, and adjust the camera toward the calibration table so that the camera is aimed at the predetermined bridge key component measurement point. Determine the camera's exposure time, aperture size, contrast and other parameter settings according to the on-site environment, and ultimately achieve high-quality image acquisition and high-frequency displacement monitoring at the hundred-meter level. Equip the monitoring station with external wires, guardrails and other ancillary facilities to complete the equipment deployment.

[0076] Image acquisition: When the power is turned on, the image synchronization acquisition module triggers the high and low levels at regular intervals to start multi-camera synchronous image acquisition.

[0077] Image preprocessing: The image and data processing module performs filtering (Gaussian filtering / stop filtering), binarization, morphological processing (erosion / dilation) and other operations on the collected images to remove noise in the image and enhance the effective feature information in the image.

[0078] Displacement data extraction: Based on the actual size of the target captured at a long distance and the indoor camera calibration results, the image distortion after preprocessing is removed, and the pixel coordinates and real space coordinates are accurately converted. If the target structure image background is complex, the convolutional neural network Flownet is selected to complete the displacement field prediction between the two input frames; if the target structure background is simple, the Lucas-Kanade method is used to estimate the sparse optical flow and realize the displacement extraction of feature points.

[0079] Data feedback: All displacement time-course data are regularly transmitted to the bridge maintenance management platform through the wireless transmission module for unified management.

[0080] Twin analysis of the overall displacement behavior of bridges: Comprehensively use statistical analysis, time series analysis, machine learning and other methods to study the spatial distribution characteristics and temporal evolution characteristics of the displacement data of key components, realize the fusion and comprehensive utilization of displacement monitoring data, and provide a data information basis for subsequent more comprehensive twin status assessment.

[0081] (1) Time domain feature analysis: Calculate the statistical characteristics of the displacement data of each component such as the main beam, sling, support, etc., including the mean, standard deviation, skewness, kurtosis, etc. This part is implemented using methods in the prior art and is not protected by the present invention. The statistical characteristics of the displacement data of each component such as the main beam, sling, support, etc., and the calculation methods of the mean, standard deviation, skewness, and kurtosis are all applicable to the present invention.

[0082] (2) Frequency domain feature analysis: The displacement data is converted from the time domain to the frequency domain through Fourier transform (FFT). This part is implemented using methods in the prior art and is not protected by the present invention. The calculation methods for extracting the frequency components, power spectrum density and other characteristics of components such as main beams and slings are applicable to the present invention.

[0083] (3) Abnormal state identification: Based on the inherent change law of structural displacement under different external conditions, the displacement response model under different working conditions is established. The statistical method Z-score is used to calculate and analyze the change trend of the time domain and frequency domain of the monitoring displacement data, identify abnormal displacement patterns, and discover potential faults in time. The following formula is used as an example for analysis of displacement time history data:

[0084]

[0085] in, and They are i The mean and standard deviation of the displacement of each component are , then it is considered that the component is at time t An exception occurs. θ is the preset threshold, otherwise, the component t No abnormalities were observed.

[0086] (4) Spatial correlation analysis of bridge components: Calculate the spatial correlation between different components to reveal their mutual influence in space and gain a more comprehensive understanding of the overall performance and behavior patterns of the bridge. This will further study the coordinated failure mechanism of components such as suspenders under different working conditions and analyze their interaction and influence under dynamic loads.

[0087]

[0088] in, and They are i and j Components at time t The displacement of and They are i and j The average displacement of the components, T is the total number of time steps, is the spatial correlation between the i-th component and the j-th component.

[0089] (5) Causal analysis of bridge overall displacement: Analyze the causes of displacement changes of different components by using causal inference methods, identify key influencing factors, and determine whether a component displacement time series is helpful in predicting another time series based on Granger causality test. Determining whether a component displacement time series is helpful in predicting another time series based on Granger causality test includes: (a) Testing the displacement time series of the i-th component and the jth component displacement time series ;

[0090] (b) Setting the null hypothesis : displacement of the ith component Not the jth component displacement Granger reasons;

[0091] (c) Construct a simplified model for displacement prediction:

[0092]

[0093] in, is the jth component at time t The predicted displacement value of is the intercept, is the autoregressive coefficient, is the displacement monitoring value of the jth component at time t-1, t-2, …, tp, is the error term;

[0094] (d) Construct a complete model for displacement prediction:

[0095]

[0096] in, is the jth component at time t The predicted displacement value of is the intercept, is the regression coefficient, is the displacement monitoring value of the jth component at time t-1, t-2, …, tp, is the displacement monitoring value of the i-th component at time t-1, t-2, …, tp, is the error term;

[0097] (e) Calculate the F statistic to test the null hypothesis:

[0098]

[0099] in, and are the residual sums of squares of the reduced model and the full model, p is the number of additional lags, T is the sample size, k is the number of parameters of the full model;

[0100] (f) Based on the distribution of the F-statistic, the corresponding probability value is calculated using the F-distribution table, and the probability value is used to determine whether a component displacement time series is helpful in predicting another time series.

[0101] If the probability value x is less than 0.05, the original hypothesis is rejected and the displacement of the i-th component is considered is the displacement of the jth component Granger reasons;

[0102] If the probability value x ≥ 0.05, the null hypothesis cannot be rejected and the displacement of the i-th component is considered Not the jth component displacement Granger reasons.

[0103] (6) Prediction of the evolution of the overall displacement of the bridge: Based on the ARIMA time series model, the changing trend of the displacement data is predicted.

[0104] The change trends of the predicted displacement data include:

[0105]

[0106] in, is a point in time t The predicted displacement value of is the intercept, is the autoregressive coefficient, is the displacement monitoring value of the point at time t-1, t-2, ..., tp or the maximum and average displacement values ​​in the nearby interval, is the error term.

[0107] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A dynamic visual bridge overall displacement monitoring and behavior twin analysis method, characterized in that: The following steps are involved: Step 1, obtaining bridge component displacement monitoring data based on the bridge component image; Step 2: Analyze the time domain characteristics and frequency domain characteristics of the displacement monitoring data; Step 3: Based on the inherent variation law of bridge structure displacement under different external conditions, a displacement response model under different working conditions is established, and the variation trend of displacement monitoring data in time domain and frequency domain is analyzed based on the Z-score statistical method to identify abnormal displacement patterns; Step 4: According to i and j Components at time t The displacement and average displacement are calculated to calculate the spatial correlation between different components in order to analyze the interaction and influence of different components under dynamic loads; Step 5: Based on the ARIMA time series model and the abnormal displacement pattern components t -1. t -2, ..., t - p The displacement monitoring data at the moment or the maximum and average displacement values ​​in the nearby interval are used to predict the change trend of the bridge displacement data; Step six: Use causal inference methods to determine whether a component displacement time series can help predict another time series, and use this to analyze the causes of displacement changes of different components and identify key influencing factors.

2. According to claim 1, a dynamic visual bridge overall displacement monitoring and behavior twin analysis method is characterized in that: Step three also includes: constructing a change trend model of the displacement time history data in the time domain and frequency domain: in, and They are i The mean and standard deviation of the displacement of each component are , then it is considered that the component is at time t An exception occurs. θ is the preset threshold, otherwise, the component t No abnormalities were observed.

3. According to claim 1, a dynamic visual bridge overall displacement monitoring and behavior twin analysis method is characterized in that: In step 4, the calculation of the spatial correlation between different components includes: in, and They are i and j Components at time t The displacement of and They are i and j The average displacement of the components, T is the total number of time steps, is the spatial correlation between the i-th component and the j-th component.

4. According to claim 1, a dynamic visual bridge overall displacement monitoring and behavior twin analysis method is characterized in that: In step 5, the change trend of the predicted displacement data includes: in, is a point in time t The displacement prediction value, 𝛼 is the intercept, is the autoregressive coefficient, That point is t -1. t -2, ..., t - p The displacement monitoring value at the moment or the maximum and average displacement values ​​in the nearby interval, is the error term.

5. According to claim 1, a dynamic visual bridge overall displacement monitoring and behavior twin analysis method is characterized in that: In step six, the causal inference method is used to analyze the causes of displacement changes of different components, identify key influencing factors, and determine whether a component displacement time series can help predict another time series based on the Granger causality test.

6. The method of dynamic visual bridge overall displacement monitoring and behavior twin analysis according to claim 5 is characterized in that: The Granger causality test is used to determine whether a component displacement time series is helpful in predicting another time series. S51 Test the displacement time series of the i-th component and the jth component displacement time series : S52 sets the null hypothesis : displacement of the ith component Not the jth component displacement Granger reasons; S53 constructs a simplified model for displacement prediction: in, is the jth component at time t The displacement prediction value, 𝛼 is the intercept, is the autoregressive coefficient, is the jth component in t -1. t -2, ..., t - p The displacement monitoring value at the moment, is the error term; S54 builds a complete model for displacement prediction: in, is the jth component at time t The displacement prediction value, 𝛼 is the intercept, is the regression coefficient, is the jth component in t -1. t -2, ..., t - p The displacement monitoring value at the moment, is the i-th component in t -1. t -2, ..., t - p The displacement monitoring value at the moment, is the error term; S55 calculates the F statistic to test the null hypothesis: in, and are the residual sums of squares of the reduced model and the full model, p is the number of additional lags, T is the sample size, k is the number of parameters of the full model; S56 then uses the F distribution table to calculate and look up the corresponding probability value based on the distribution of the F statistic value, and determines whether a component displacement time series is helpful in predicting another time series based on the probability value.

7. The method of dynamic visual bridge overall displacement monitoring and behavior twin analysis according to claim 6 is characterized in that: Step (56) further includes: If the probability value x is less than 0.05, the original hypothesis is rejected and the displacement of the i-th component is considered is the displacement of the jth component Granger reasons; If the probability value x ≥ 0.05, the null hypothesis cannot be rejected and the displacement of the i-th component is considered Not the jth component displacement Granger reasons.

8. A dynamic visual bridge overall displacement monitoring and behavior twin analysis system, characterized in that: include: A first main control module acquires bridge component displacement monitoring data based on the bridge component image; A second main control module performs time domain and frequency domain feature analysis on the displacement monitoring data; The third main control module establishes the displacement response model under different working conditions based on the inherent variation law of bridge structure displacement under different external conditions, and analyzes the variation trend of displacement monitoring data in time domain and frequency domain based on the Z-score statistical method to identify abnormal displacement patterns; The fourth main control module, according to i and j Components at time t The displacement and average displacement are calculated to calculate the spatial correlation between different components in order to analyze the interaction and influence of different components under dynamic loads; The fifth main control module is based on the ARIMA time series model and the abnormal displacement pattern component. t -1. t -2, ..., t - p The displacement monitoring data at the moment or the maximum and average displacement values ​​in the nearby interval are used to predict the change trend of the bridge displacement data; The sixth main control module uses the causal inference method to determine whether a component displacement time series is helpful in predicting another time series, and thereby analyzes the causes of displacement changes of different components and identifies key influencing factors.

9. The dynamic visual bridge overall displacement monitoring and behavior twin analysis system according to claim 8 is characterized in that: The first main control module includes: multiple cameras for acquiring images of bridge components; An image acquisition unit, used for receiving images acquired by a camera; The image and data processing unit is used to process images to obtain bridge component displacement monitoring data.

10. The dynamic visual bridge overall displacement monitoring and behavior twin analysis system according to claim 9 is characterized in that: The first main control module also includes: The camera orientation calibration platform is used to calibrate the camera orientation so that a complete component image can be obtained by cooperating with multiple cameras; The image synchronization acquisition module is used to synchronize the working time of all cameras.

Images