Bridge Dynamic Weighing and Graded Alarm Method Based on Visual Strain Monitoring

Abstract

The present invention discloses a method for dynamic weighing and hierarchical alarm of bridges based on visual strain monitoring, belonging to the technical field of vehicle weighing; the method for dynamic weighing and hierarchical alarm of bridges based on visual strain monitoring includes: identifying vehicle parameters on the bridge, where the vehicle parameters include: the number of axles, wheelbase, vehicle speed, and the number of vehicles; monitoring the strain at the mid-span of the bridge; based on the linear relationship between vehicle axle weight and the strain at the mid-span of the bridge, as well as vehicle parameters, establishing a linear regression model considering uncertainty, and based on Bayesian inference, identifying the axle weight parameters of the vehicle and their confidence ranges, and finally obtaining the total vehicle weight and its confidence range; determining the weight limit standard based on the number of axles, and then combining with the dynamic weighing result to perform overweight hierarchical alarm; thereby improving the accuracy and lifespan of the monitoring system, reducing the maintenance cost of the bridge dynamic weighing system, and effectively avoiding damage or even destruction of major bridges caused by overloaded vehicles.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle weighing, and in particular relates to a bridge dynamic weighing and graded alarm method based on visual strain monitoring. Background Art

[0002] Vehicle loads are a significant component of bridges. Excessive vehicle loads can severely damage bridge infrastructure, shortening its service life and even exceeding its load capacity, leading to bridge overturning or collapse. Therefore, identifying vehicle loads and providing alerts for overweight vehicles is crucial to the health of bridge infrastructure.

[0003] The current vehicle weighing methods mainly include road-type weighing methods and bridge dynamic weighing methods. Among them, road-type weighing methods include traditional floor-scale static weighing methods and road-type dynamic weighing methods. The traditional floor-scale static weighing method requires a special weighing station, and the vehicle needs to be parked on the floor scale for weighing. The recognition efficiency is low and the weighing cost is high. The road-type dynamic weighing method achieves the purpose of dynamic weighing by installing sensors on the road surface and monitoring the dynamic response of the road surface when the vehicle passes through the sensor. The recognition efficiency of this weighing method is significantly higher than that of the traditional floor-scale static weighing method, but it requires the weighing system to be installed in the groove of the road surface. The cost of system installation and maintenance is high, and the recognition accuracy and stability are not very high.

[0004] In recent years, dynamic weighing methods for bridges have been continuously proposed. They identify the weight of vehicles by monitoring the dynamic response of vehicles passing through bridges. This weighing method does not affect the normal passage of vehicles, does not require the installation of large equipment such as scales, and has improved weighing accuracy. The dynamic response of bridges monitored by this weighing method can also be used to evaluate the service performance and health status of bridges.

[0005] However, most existing bridge weigh-in-motion methods require the installation of various sensors on or under the bridge. These sensors are in direct contact with the bridge structure, significantly affecting their performance and accuracy due to bridge vibrations. This increases weighing system maintenance costs and reduces sensor lifespan. Due to the complexity of vehicle-bridge coupled vibrations, the ill-posedness of the inverse dynamics problem, and noise interference from response monitoring, the uncertainty of vehicle weights identified by bridge weigh-in-motion methods is high, making this method still far from practical application. Summary of the Invention

[0006] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a bridge dynamic weighing and graded alarm method based on visual strain monitoring.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] The bridge dynamic weighing method based on visual strain monitoring includes the following steps:

[0009] Identify the vehicle parameters on the bridge. The vehicle parameters include: the number of axles, the wheelbase, the vehicle speed, and the number of vehicles.

[0010] Monitor the strain at the mid-span of the bridge.

[0011] Based on the linear relationship between the vehicle axle load and the mid-span strain of the bridge, as well as the vehicle parameters, establish a linear regression model considering uncertainty, and identify the axle load parameters and their confidence ranges of the vehicle based on Bayesian inference. Finally, obtain the total vehicle weight and its confidence range.

[0012] Furthermore, two groups of laser distance sensors and baffles are arranged on both sides of the lane at the bridge support end. The two groups of laser distance sensors and baffles are respectively arranged at the entrance end and the exit end of the bridge, and they are respectively located outside the lane lines on both sides of the driving lane. The connection line is horizontal and perpendicular to the lane line, and the installation height is lower than the minimum ground clearance of highway vehicles. Use the laser distance sensors to monitor and record the distance from the sensors to the baffles or the wheels of the passing vehicles, and then identify the vehicle parameters according to the characteristics of the distance time-history parameters monitored by the laser distance sensors.

[0013] Furthermore, the identification steps of the vehicle parameters include:

[0014] S11. Based on the distance time-history data D1(t) and D2(t) monitored by the laser distance sensors at the entrance end and the exit end of the bridge, with the lane width D0 as the reference, calculate the negative interval numbers N1 and N2 of the functions (D1(t) - 0.5D0) and (D2(t) - 0.5D0). Each negative interval corresponds to an axle, and the number of axles N on the bridge a is:

[0015] N a = N1 - N2;

[0016] S12. The speed V of the i-th axle i is:

[0017] V i = L / (T2 i - T1 i )

[0018] where L is the bridge span, and T1 i and T2 i are respectively the moments when the i-th negative interval appears after subtracting 0.5D0 from the distance time-histories at the entrance end and the exit end of the bridge;

[0019] S13. The wheelbase D between the i-th axle and the (i + 1)-th axle a is:

[0020] D a = (T1i+1 -T1 i )×V i =(T2 i+1 -T2 i )×V i ;

[0021] S14, Cluster the speeds of all axles on the bridge. Axles with the same speed belong to the same vehicle, thereby determining the number of vehicles on the bridge and the axles corresponding to each vehicle.

[0022] Further, the steps of monitoring the strain at the mid-span of the bridge include:

[0023] S21, Arrange a uniformly distributed group of marking points at the bottom of the beam at the mid-span of the bridge, and set up a binocular camera directly below the group of marking points and perform parameter calibration;

[0024] S22, When the laser range sensor at the entrance end of the bridge recognizes that a vehicle enters the bridge, trigger the binocular camera to take pictures of the group of marking points at a fixed frequency to obtain a sequence of images;

[0025] S23, Use the feature matching method to measure the stereo parallax of the sequence of images, and then determine the three-dimensional coordinates of the group of marking points based on the mapping relationship between the image coordinate system and the bridge coordinate system;

[0026] S24, Monitor the strain time history at the mid-span of the bridge based on the Green-Lagrange strain.

[0027] Further, the method for arranging the group of marking points is: Arrange at least 5 marking points on both sides of the mid-span center line at the bottom of the bridge beam. The marking points are symmetrically distributed about the mid-span center line and the distance from the center line does not exceed 5 cm. The marking points are uniformly distributed along the transverse direction of the bridge.

[0028] Further, in S24, based on the Green-Lagrange strain, determine the longitudinal strain ε at each position at the mid-span of the bridge xi , specifically:

[0029]

[0030] In the formula, f is the monitoring frequency; t is the time; x ri (t) and x li (t) are the longitudinal coordinates of the marking points on both sides of the strain monitoring position in the bridge coordinate system, respectively.

[0031] Further, the steps of the dynamic weighing include:

[0032] S31, Establish a linear regression model according to the linear relationship between the axle weight of the vehicle and the strain at the mid-span of the bridge, specifically:

[0033] R = IA + ε

[0034] Wherein, R is a vector composed of the measured bridge mid-span strains obtained by S24; A is the axle weight vector to be identified; I is the influence coefficient matrix; ε is the error term vector, which satisfies a Gaussian distribution with a mean of 0 and a variance of σ 2 λ Vi ; λ is the error parameter;

[0035] S32. Determine the posterior distributions of the axle weight vector and the error parameter according to Bayesian inference, and obtain the optimal value equations of each parameter by maximizing the posterior distribution, which are:

[0036] A(σ 2 ,λ)=(σ -2 λ -V I T I + E) -1 (σ -2 λ -V I T R + 10E)

[0037]

[0038] Wherein: E is the identity matrix; N is the number of strain acquisitions; α0 and β0 are the shape and scale parameters of the variance distribution of the error term respectively;

[0039] S33. Iteratively solve the optimal value equations to obtain the conditional optimal values and posterior distributions of each parameter, and then determine the axle weight parameters of the vehicle and its confidence range at the α confidence level as Σ is the covariance of the posterior distribution of the axle weight parameters;

[0040] S34. Combine the correspondence between the vehicles and axles on the bridge, and sum to obtain the total vehicle weight and its confidence range.

[0041] The bridge dynamic weighing system based on visual strain monitoring includes:

[0042] Parameter identification module: Identify the vehicle parameters on the bridge, and the vehicle parameters include: the number of axles, the wheelbase, the vehicle speed, and the number of vehicles;

[0043] Strain monitoring module: Monitor the bridge mid-span strain;

[0044] And, dynamic weighing module: Establish a linear regression model considering uncertainty based on the linear relationship between the vehicle axle weight and the bridge mid-span strain, and the vehicle parameters, and identify the axle weight parameters of the vehicle and their confidence range based on Bayesian inference, and finally obtain the total vehicle weight and its confidence range.

[0045] The bridge grading alarm method based on visual strain monitoring includes the above weighing method, and further includes: Determine the weight limit standard based on the number of axles, and then combine the dynamic weighing results to perform overweight grading alarm.

[0046] Furthermore, the steps of overweight grading alarm include:

[0047] S41. Determine the corresponding weight limit standard according to the number of axles of the vehicle on the bridge;

[0048] S42. If the total weight of the vehicle exceeds the weight limit standard by 100%, trigger a first-level alarm;

[0049] S43. If the total weight of the vehicle exceeds the weight limit standard by 20% and does not reach 100%, trigger a second-level alarm;

[0050] S44. If the total weight of the vehicle exceeds the weight limit standard but does not reach 20%, trigger a third-level alarm;

[0051] S45. If the total weight of the vehicle does not exceed the weight limit standard, no alarm is triggered and the vehicle can pass the bridge normally.

[0052] Advantages of the present invention:

[0053] 1. The present invention realizes the dynamic weighing of vehicles on the bridge through the identification of vehicle parameters on the bridge and the monitoring of the mid-span strain of the bridge, and conducts grading alarm, which can effectively alarm overweight vehicles in real time, helps the relevant management departments to judge and evaluate overweight vehicles in time, and effectively avoids damage or even destruction of major bridges caused by overweight vehicles.

[0054] 2. The present invention monitors the mid-span strain response of the bridge by means of vision technology, which belongs to non-contact strain monitoring means, solves the problems of easy damage of contact monitoring sensors and unstable monitoring data, improves the accuracy and service life of the monitoring system, and reduces the maintenance cost of the bridge dynamic weighing system.

[0055] 3. The present invention considers the uncertainties in axle weight dynamic identification caused by factors such as bridge dynamic response, road unevenness, and environmental interference, and identifies the axle weight parameters and their confidence ranges of the vehicle based on Bayesian inference, and can determine the identified vehicle weight results at different confidence levels, improving the accuracy of vehicle load identification on the bridge and effectively avoiding misjudgment of vehicle overweight. Description of the drawings

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0057] Figure 1 It is a flow chart of the bridge dynamic weighing and grading alarm method of the present invention;

[0058] Figure 2Schematic diagram of the layout of devices such as laser ranging sensors and binocular cameras;

[0059] Figure 3 Graph of the distance time history data monitored by the laser ranging sensor;

[0060] Figure 4 Layout diagram of the marking point group at the bottom of the beam at the mid-span of the bridge;

[0061] Figure 5 Parallax diagram of the marking point group obtained by visual stereo matching;

[0062] Figure 6 Flowchart for realizing vehicle overweight classification alarm. Detailed implementation manners

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0064] As Figure 1 shown, the bridge dynamic weighing and classification alarm method based on visual strain monitoring includes the following steps:

[0065] S1. Identify vehicle parameters on the bridge. The vehicle parameters include: the number of axles, wheelbase, vehicle speed, and the number of vehicles;

[0066] As Figure 2 shown, two groups of laser ranging sensors and baffles are arranged on both sides of the lane at the bridge support end. The two groups of laser ranging sensors and baffles are respectively arranged at the entrance end and the exit end of the bridge, and are respectively located outside the two lane lines of the driving lane. The connection line is horizontal and perpendicular to the lane line, and the arrangement height is lower than the minimum clearance between the vehicle chassis and the ground of highway vehicles; the laser ranging sensor is used to monitor and record the distance from the sensor to the baffle or the wheels of the passing vehicle, and then the number of axles, wheelbase, vehicle speed, and the number of vehicles of the vehicle are identified according to the distance time history parameter characteristics monitored by the laser ranging sensor;

[0067] Among them, the specific steps of parameter identification include:

[0068] S11. Based on the distance time history data D1(t) and D2(t) monitored by the laser ranging sensors at the entrance end and the exit end of the bridge, as Figure 3 shown, taking the lane width D0 as a reference, calculate the negative interval numbers N1 and N2 of the functions (D1(t) - 0.5D0) and (D2(t) - 0.5D0). Each negative interval corresponds to an axle, and the number of axles N a on the bridge is:

[0069] N a = N1 - N2;

[0070] S12, the speed V of the i-th axle i is:

[0071] V i = L / (T2 i - T1 i )

[0072] where L is the bridge span, T1 i and T2 i are respectively the moments when the i-th negative interval appears after subtracting 0.5D0 from the time histories of the corresponding distances at the entrance and exit ends of the bridge;

[0073] S13, the wheelbase D between the i-th axle and the (i + 1)-th axle a is:

[0074] D a = (T1 i+1 - T1 i ) × V i = (T2 i+1 - T2 i ) × V i ;

[0075] S14, perform clustering analysis on the speeds of all axles on the bridge. Axles with the same speed belong to the same vehicle, and the speed of the vehicle is the speed of the axles belonging to it. Thus, determine the number of vehicles on the bridge and the axles corresponding to each vehicle.

[0076] S2, monitor the strain at the mid-span of the bridge;

[0077] Spray and arrange a uniformly distributed group of marked points on the bottom of the beam at the mid-span of the bridge with black matte paint. Set up a binocular camera directly below the group of marked points through a right-angle steel vertical rod and perform parameter calibration; when the laser range finder arranged at the entrance end of the bridge in S1 recognizes that a vehicle enters the bridge, trigger the binocular camera to take pictures of the group of marked points at a fixed frequency to obtain a sequence of images; use the feature matching method to measure the stereo parallax of the sequence of images, and then determine the three-dimensional coordinates of the group of marked points based on the mapping relationship between the image coordinate system and the bridge coordinate system; monitor the strain time history at the mid-span of the bridge based on the Green-Lagrange strain.

[0078] As Figure 4 shown, in this embodiment, the arrangement method of the group of marked points is: arrange at least 5 upper marked points on both sides of the mid-span centerline at the bottom of the bridge beam. The marked points are symmetrically distributed about the mid-span centerline and the distance from the centerline does not exceed 5 cm. The marked points are uniformly distributed along the transverse direction of the bridge;

[0079] ​The binocular camera is fixedly arranged directly below the marker point group at the bottom of the bridge beam through a right-angle steel vertical rod, and the other end of the right-angle steel vertical rod is fixed to the ground near the pier; the lens of the binocular camera faces vertically upward, and the shooting area is slightly larger than the range of the marker point group.

[0080] Among them, the specific steps for strain monitoring at the mid-span of the bridge are as follows:

[0081] S21, Calibrate the parameters of the arranged binocular camera to determine the camera internal parameter matrix and the mapping relationship matrix between the image coordinate system and the bridge coordinate system;

[0082] S22, The binocular camera shoots the marker point group at a fixed frequency and obtains a sequence of images. Rectify the epipolar lines of the left and right view images, and use the normalized cross-correlation method to obtain a disparity map with the same size as the original image, as Figure 5 shown, and then obtain the depth information of the sequence of images according to the camera internal parameter matrix;

[0083] S23, Identify the marker point group in the sequence of images through cross-matching, and then determine the three-dimensional coordinates of the marker point group based on the mapping relationship between the image coordinate system and the bridge coordinate system determined in S21 and the image depth parameters obtained in S22;

[0084] S24, Determine the longitudinal strain ε at each position in the mid-span of the bridge based on the Green-Lagrange strain xi , specifically:

[0085]

[0086] In the formula, f is the monitoring frequency; t is the time; x ri (t) and x li (t) are the longitudinal coordinates of the marker points on both sides of the strain monitoring position in the bridge coordinate system along the bridge direction, respectively.

[0087] S3, Dynamic weighing of the bridge;

[0088] According to the linear relationship between the vehicle axle weight and the mid-span strain of the bridge, as well as the vehicle parameters, establish a linear regression model considering uncertainty, and identify the axle weight parameters of the vehicle and their confidence ranges based on Bayesian inference. Finally, obtain the total vehicle weight and its confidence range;

[0089] The steps of dynamic weighing include:

[0090] S31, Considering the uncertainty in axle weight identification caused by factors such as bridge dynamic response, road unevenness, and environmental interference, establish a linear regression model according to the linear relationship between the vehicle axle weight and the mid-span strain of the bridge, specifically:

[0091] R = IA + ε

[0092] Wherein, R is a vector composed of the measured mid-span strain of the bridge obtained by S24; A is the axle weight vector to be identified; I is the influence coefficient matrix, which is determined according to the axle position, and the axle position is determined by the vehicle parameters identified in S1; ε is the error term vector, which satisfies a Gaussian distribution with a mean of 0 and a variance of σ 2 a Gaussian distribution of λVi; λ is the error parameter.

[0093] S32. Determine the posterior distributions of the axle weight vector and the error parameter according to Bayesian inference, and obtain the optimal value equation of each parameter by maximizing the posterior distribution, which is:

[0094] A(σ 2 , λ) = (σ -2 λ -V I T I + E) -1 (σ - 2λ -V I T R + 10E)

[0095]

[0096] Wherein: E is the identity matrix; N is the number of strain acquisitions; α0 and β0 are respectively the shape and scale parameters of the variance distribution of the error term.

[0097] S33. Iteratively solve the optimal value equation to obtain the conditional optimal values and posterior distributions of each parameter, and then determine the axle weight parameter of the vehicle and its confidence range at the α confidence level as Σ is the covariance of the posterior distribution of the axle weight parameter.

[0098] S34. Combine the correspondence between the vehicle on the bridge and the axles determined in S14, and sum to obtain the total vehicle weight and its confidence range.

[0099] S4. Overweight classification and alarm;

[0100] Based on the number of vehicle axles in S1, determine the weight limit standard, and then combine the dynamic weighing results in S3 to perform overweight classification and alarm;

[0101] As Figure 6 shown, the steps of overweight classification and alarm include:

[0102] S41. Determine the corresponding weight limit standard according to the number of vehicle axles on the bridge determined in S14;

[0103] S42. If the total vehicle weight exceeds the weight limit standard by 100%, trigger a first-level alarm, warning the driver to get off the bridge and notifying the traffic management department and the public security department for severe punishment;

[0104] S43, if the total vehicle weight exceeds the weight limit standard by 20% but is less than 100%, a secondary alarm is triggered to warn the driver to get off the bridge and notify the relevant management department for moderate punishment;

[0105] S44, if the total vehicle weight exceeds the weight limit standard but is less than 20%, a tertiary alarm is triggered to warn the driver to get off the bridge and impose minor punishments such as deduction of points and fines;

[0106] S45, if the total vehicle weight does not exceed the weight limit standard, no alarm is triggered and the vehicle can pass through the bridge normally.

[0107] In the description of this specification, the description with reference to terms such as "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0108] The foregoing shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and all such changes and improvements fall within the scope of the present invention claimed.

Claims

1. A bridge dynamic weighing method based on visual strain monitoring, characterized in that, Including the following steps: Identifying vehicle parameters on the bridge, where the vehicle parameters include: the number of axles, wheelbase, vehicle speed, and the number of vehicles; Monitoring the mid-span strain of the bridge; Based on the linear relationship between vehicle axle load and mid-span strain of the bridge, as well as vehicle parameters, establishing a linear regression model considering uncertainty, and identifying the axle load parameters of the vehicle and their confidence ranges based on Bayesian inference, and finally obtaining the total vehicle weight and its confidence range; The steps of monitoring the mid-span strain of the bridge include: S21, arranging a uniformly distributed group of marked points at the bottom of the beam at the mid-span of the bridge, and setting up a binocular camera directly below the group of marked points and calibrating its parameters; S22, when the laser range sensor at the entrance end of the bridge identifies a vehicle entering the bridge, triggering the binocular camera to take pictures of the group of marked points at a fixed frequency to obtain a sequence of images; S23, using the feature matching method to measure the stereo disparity of the sequence of images, and then determining the three-dimensional coordinates of the group of marked points based on the mapping relationship between the image coordinate system and the bridge coordinate system; S24, monitoring the strain time history at the mid-span of the bridge based on the Green-Lagrange strain; The steps of obtaining the total vehicle weight and its confidence range include: S31, establishing a linear regression model according to the linear relationship between vehicle axle load and mid-span strain of the bridge, specifically: R = IA + ε wherein, R is a vector composed of the measured strain time history at the mid-span of the bridge obtained by S24; A is the axle load vector to be identified; I is the influence coefficient matrix; ε is the error term vector, which satisfies a Gaussian distribution with a mean of 0 and a variance of σ 2 λ Vi ; λ is the error parameter; S32, determining the posterior distributions of the axle load vector and error parameters according to Bayesian inference, and obtaining the optimal value equations of each parameter by maximizing the posterior distribution, which are: A(σ 2 ,λ)=(σ -2 λ -V I T I + E) -1 (σ -2 λ -V I T R + 10E) In the formula: E is the identity matrix; N is the number of strain acquisitions; α0 and β0 are the shape and scale parameters of the variance distribution of the error term respectively; S33. Iteratively solve the optimal value equation to obtain the conditional optimal values and posterior distributions of each parameter, and then determine the axle weight parameters of the vehicle and their confidence intervals at the α confidence level as Σ is the covariance of the posterior distribution of the axle weight parameter; S34, combining the correspondence between the vehicles and axles on the bridge, and summing to obtain the total vehicle weight and its confidence range.

2. The method for dynamic weighing of bridges based on visual strain monitoring according to claim 1, wherein Two groups of laser range sensors and baffles are arranged on both sides of the lane at the bridge support end. The two groups of laser range sensors and baffles are respectively arranged at the entrance end and the exit end of the bridge, and they are respectively located outside the lane lines on both sides of the driving lane. Their connection lines are horizontal and perpendicular to the lane lines, and the installation height is lower than the minimum clearance between the highway vehicle chassis and the ground; using the laser range sensors to monitor and record the distances from the sensors to the baffles or the wheels of the passing vehicles, and then identifying the vehicle parameters according to the characteristics of the distance time history parameters monitored by the laser range sensors.

3. The method for dynamic weighing of bridges based on visual strain monitoring according to claim 2, wherein, The steps of identifying vehicle parameters on the bridge include: S11. Based on the distance time - history data D1(t) and D2(t) monitored by the laser ranging sensors at the entrance and exit ends of the bridge, taking the lane width D0 as a reference, calculate the number of negative intervals N1 and N2 of the functions (D1(t)-0.5D0) and (D2(t)-0.5D0). Each negative interval corresponds to an axle, and the number of axles N on the bridge a is as follows: N a = N1 - N2; S12, the speed V of the i-th axle i is as follows: V i = L / (T2 i - T1 i ) where L is the bridge span, T1 i and T2 i are respectively the occurrence times of the i-th negative interval after subtracting 0.5D0 from the time histories of the corresponding distances at the entrance and exit ends of the bridge; S13, wheelbase D between the i-th axle and the (i + 1)-th axle a is as follows: D a = (T1 i+1 - T1 i ) × V i = (T2 i+1 - T2 i ) × V i ; S14, performing clustering analysis on the speeds of all axles on the bridge. The axles with the same speed belong to the same vehicle, thereby determining the number of vehicles on the bridge and the axles corresponding to each vehicle.

4. The method for dynamic weighing of bridges based on visual strain monitoring according to claim 2, characterized in that, The method for arranging the group of marked points is: arranging at least 5 marked points on both sides of the mid-span center line at the bottom of the bridge beam. The marked points are symmetrically distributed about the mid-span center line and the distance from the center line does not exceed 5 cm, and the marked points are evenly distributed along the transverse direction of the bridge.

5. The method for dynamic weighing of bridges based on visual strain monitoring according to claim 2, characterized in that In S24, the longitudinal strain ε at each position in the mid-span of the bridge is determined based on the Green-Lagrange strain xi , specifically: Where f is the monitoring frequency; t is the time; x ri (t) and x li (t) are the longitudinal coordinates of the marker points on both sides of the strain monitoring position in the bridge coordinate system, respectively.

6. Bridge dynamic weighing system based on visual strain monitoring, characterized in that, Including: Parameter identification module: identifying vehicle parameters on the bridge, where the vehicle parameters include: the number of axles, wheelbase, vehicle speed, and the number of vehicles; Strain monitoring module: monitoring the mid-span strain of the bridge; And a dynamic weighing module: based on the linear relationship between vehicle axle load and mid-span strain of the bridge, as well as vehicle parameters, establishing a linear regression model considering uncertainty, and identifying the axle load parameters of the vehicle and their confidence ranges based on Bayesian inference, and finally obtaining the total vehicle weight and its confidence range; The steps of monitoring the mid-span strain of the bridge include: S21. Arrange a uniformly distributed group of marking points at the bottom of the beam at the mid-span of the bridge, and set up a binocular camera directly below the group of marking points and calibrate its parameters. S22. When the laser range finder at the entrance end of the bridge identifies that a vehicle enters the bridge, trigger the binocular camera to capture the group of marking points at a fixed frequency to obtain a sequence of images. S23. Use the feature matching method to measure the stereo parallax of the sequence of images, and then determine the three-dimensional coordinates of the group of marking points based on the mapping relationship between the image coordinate system and the bridge coordinate system. S24. Monitor the strain time history at the mid-span of the bridge based on the Green-Lagrange strain. The steps of obtaining the total vehicle weight and its confidence range include: S31. Establish a linear regression model according to the linear relationship between the axle weight of the vehicle and the strain at the mid-span of the bridge, specifically: R = IA + ε wherein, R is a vector composed of the measured strain time history at the mid-span of the bridge obtained by S24; A is the axle weight vector to be identified; I is the influence coefficient matrix; ε is the error term vector, which satisfies the Gaussian distribution with a mean of 0 and a variance of σ 2 λ Vi ; λ is the error parameter; S32. Determine the posterior distribution of the axle weight vector and the error parameter according to Bayesian inference, and obtain the optimal value equation of each parameter by maximizing the posterior distribution, which is: A(σ 2 ,λ) = (σ -2 λ -V I T I + E) -1 (σ -2 λ -V I T R + 10E) In the formula: E is the identity matrix; N is the number of strain acquisitions; α0 and β0 are the shape and scale parameters of the variance distribution of the error term respectively. S33. Iteratively solve the optimal value equation to obtain the conditional optimal values and posterior distributions of each parameter, and then determine the axle weight parameter of the vehicle and its confidence range at the α confidence level as Σ is the covariance of the posterior distribution of the axle weight parameter; S34. Combine the correspondence between the vehicle on the bridge and the axles, and sum to obtain the total vehicle weight and its confidence range.

7. A bridge hierarchical alarm method based on visual strain monitoring, characterized in that, It includes the weighing method according to any one of claims 1-5, and further includes: determining the weight limit standard based on the number of axles, and then combining the dynamic weighing result to perform overweight classification alarm.

8. The bridge grading alarm method based on visual strain monitoring according to claim 7, wherein The steps of the overweight classification alarm include: S41. Determine the corresponding weight limit standard according to the number of axles of the vehicle on the bridge. S42. If the total vehicle weight exceeds the weight limit standard by 100%, trigger a first-level alarm. S43. If the total vehicle weight exceeds the weight limit standard by 20% and does not reach 100%, trigger a second-level alarm. S44. If the total vehicle weight exceeds the weight limit standard but does not reach 20%, trigger a third-level alarm. S45. If the total vehicle weight does not exceed the weight limit standard, no alarm is triggered and the vehicle can pass through the bridge normally.

Images