Bridge group spatio-temporal load distribution identification method based on multi-source data fusion

Abstract

The present application belongs to the technical field of bridge group load distribution identification. Disclosed is a bridge group spatio-temporal load distribution identification method based on multi-source data fusion. The problem in the prior art of traditional bridge group load distribution identification methods requiring an additional device and having difficulties in identifying the spatio-temporal distribution of vehicles on a bridge at night is solved. In the present application, on the basis of the vehicle correlation and time correlation between vehicle positioning data and vehicle load detection data in combination with the feature that the load of a vehicle remains unchanged within a certain time, by using a vehicle load detection point as a node, a vehicle trajectory is divided into a plurality of load sections, and a load section where a bridge is located is acquired; moreover, on the basis of a vehicle ID and a vehicle load point detection time, vehicle positioning data in the load section where the bridge is located is associated and matched with vehicle load detection point data to acquire actual load data when the vehicle passes over the bridge. The present application can be applied on a large scale and throughout the day to acquire a vehicle load closer to a true value, requires a lower cost and can be applied to the detection of bridge group loads.

Description

A method for identifying spatiotemporal load distribution of bridge groups based on multi-source data fusion Technical Field

[0001] The present application relates to a method for identifying the spatiotemporal load distribution of a bridge group, and in particular to a method for identifying the spatiotemporal load distribution of a bridge group based on multi-source data fusion, belonging to the technical field of bridge group load distribution identification. Background Art

[0002] Vehicle loads are the primary external loads borne by road and bridge structures. These loads are subject to significant uncertainty, influenced by travel demand, the environment, traffic control, and the bridge's location. Furthermore, the varying vehicle types, weights, axle loads, and speeds on bridges complicates the determination of vehicle load distribution. Inaccurate vehicle load distribution leads to inconsistencies between vehicle load models and the actual loads borne by the bridge. This uncertainty in load input makes it difficult to assess bridge safety based on structural response monitoring data. Therefore, obtaining accurate spatiotemporal load distribution on bridges is crucial for developing vehicle load distribution models and for studying the interplay between bridge performance and traffic loads.

[0003] Currently, there are two main bridge vehicle load detection technologies: one is based on a dynamic weighing system to accurately measure the weight and speed of vehicles passing on the bridge deck, and obtain the bridge deck vehicle load probability distribution model through statistical analysis methods or by arranging multiple cameras on the bridge deck to obtain the spatial distribution of vehicles through imaging methods. However, it can only be used for the bridge deck load distribution of a single bridge, and is costly, time-consuming, and complex to operate. The other is based on the monitoring system structural response data and the vehicle-bridge coupling model, and identifies vehicle loads through methods such as influence line methods and machine learning. However, in actual application, a structural health monitoring system needs to be installed, which is costly, time-consuming, and complex to operate.

[0004] In the prior art, the patent document with the publication (announcement) number CN108914815B discloses a bridge deck vehicle load identification device, a bridge, and a bridge load distribution identification method. The radar tracking and positioning system includes at least one radar group, and the radar group includes three radars. The radar is used to be set on the bridge to collect the first type of vehicle data. The dynamic weighing system is used to be set at intervals on the bridge deck to collect the second type of vehicle data. The data processing equipment obtains the first type of vehicle data and the second type of vehicle data. The data processing equipment calculates the vehicle driving trajectory associated with time based on the clock information and the first type of vehicle data, and combines the second type of vehicle data and the vehicle driving trajectory associated with time to obtain the spatial distribution of the vehicle load on the bridge deck. The spatial distribution of the vehicle load on the bridge deck at any time can be obtained, but it is not suitable for bridges without radar equipment, video equipment, and dynamic weighing systems. The patent document with the notification number CN111709332B discloses a method for identifying the spatiotemporal distribution of vehicle loads on bridges based on dense convolutional neural networks. Multiple cameras are installed at different positions on the bridge to obtain images on the bridge from multiple directions, and video images with time tags are output. The dense neural network is used to obtain the multi-channel characteristics of vehicles on the bridge, including color features, shape features, position features, etc. The vehicle data and features under different cameras at the same time are analyzed to obtain the vehicle distribution on the bridge at any time. The dynamic weighing system is combined to obtain the vehicle load of vehicles passing through the bridge, and the spatiotemporal distribution of vehicle loads on the bridge deck is recognized. However, it requires the installation of multiple cameras on the bridge and requires a video recognition algorithm for recognition. The operation is difficult, costly, and requires a lot of computing power. At the same time, it is difficult to identify the vehicle load distribution at night, and most heavy-loaded vehicles often travel at night.

[0005] In summary, a bridge group load distribution identification method is needed that can obtain vehicle loads throughout the day, does not require additional equipment, and can be applied on a large scale.

[0006] Application Contents

[0007] A brief overview of the present application is provided below to provide a basic understanding of certain aspects of the present application. It should be understood that this overview is not an exhaustive overview of the present application. It is not intended to identify key or important portions of the present application, nor is it intended to limit the scope of the present application. Its purpose is simply to present certain concepts in a simplified form as a prelude to the more detailed description that will be discussed later.

[0008] In view of this, in order to solve the problem that the traditional bridge group load distribution identification method in the prior art requires additional equipment and is difficult to identify the spatiotemporal distribution of vehicles on the bridge deck at night, the present application provides a bridge group spatiotemporal load distribution identification method based on multi-source data fusion.

[0009] The technical solution is as follows: A method for identifying spatiotemporal load distribution of bridge groups based on multi-source data fusion includes the following steps:

[0010] S1. Perform GIS topology network matching on vehicles based on vehicle positioning data to obtain vehicle trajectory data;

[0011] S11. Select vehicle load detection points within the area and extract and collect vehicle load detection data and vehicle positioning data;

[0012] Specifically: Vehicle load detection points include overload control points, source overload control detection points, high-speed weight toll detection points and bridge dynamic weighing detection points. Vehicle load detection data and vehicle positioning data within the same selected time period are selected in the selected area. Vehicle load detection data, that is, the collected vehicle load detection point information includes license plate number, detection time, vehicle load, and the section ID, longitude and latitude of the section where the vehicle load is located. Vehicle positioning data includes vehicle length, vehicle model, vehicle type, license plate number, time, driving speed and latitude and longitude coordinates.

[0013] S12 pre-processes the vehicle positioning data and collects GIS topological network section information of the road network;

[0014] S13. Collect bridge group information and match vehicle positioning data with GIS topological network section information;

[0015] S14. According to the GIS topology network section information, obtain the vehicle positioning data matching each bridge line section in the bridge group to form vehicle trajectory data;

[0016] S2. Segment the vehicle trajectory based on the vehicle load detection point, and obtain the path section where the bridge is located in combination with the road section where the bridge is located;

[0017] S21. Divide all vehicle trajectories into multiple path sub-segments according to the road section ID of the vehicle load detection point;

[0018] S22. According to the bridge group information, the bridge line is located in the road section, extracting the bridge group of each bridge each bridge line where the path sub-section;

[0019] S3. Based on the vehicle positioning data and vehicle load data, identify the vehicle loads matching the start and end points of each line path segment of the bridge in the bridge group and calculate the actual vehicle load passing through the bridge;

[0020] S31. For each vehicle load segment set matched to each bridge row, identify the starting and ending vehicle load detection points corresponding to the vehicle entry and exit segments of each path subsegment, and calculate the detection time and vehicle load passing through these two points;

[0021] S32. Select a specific vehicle load detection point, and calculate the predicted time of a specific vehicle load detection point based on the vehicle positioning data, the predicted time of the vehicle load detection point after the starting point, and the predicted time of the vehicle after the end point vehicle load detection point;

[0022] S33. The vehicle positioning data is matched with the vehicle load data of the section from the starting point to the end point of the vehicle load detection point, ie, the load section, to obtain the actual vehicle load;

[0023] S4. Based on the vehicle positioning data and the vehicle load detection point detection data, identify the bridge line path section starting and ending points matching the vehicle load, and obtain the vehicle load passing the bridge;

[0024] S41. Select the vehicle positioning data of the road section where the bridge is located, and construct a lane-level GIS topology network of the road section where the bridge is located;

[0025] S42. Determine the matching priority based on the sampling frequency, construct a set of vehicle positioning sequences with different matching priorities for the vehicle positioning data of the road section where the bridge is located, and calculate the probability score of the vehicle positioning data point matching each lane;

[0026] S43. Use a Gaussian distribution model for fitting, identify the possibility of vehicle lane change, calculate the vehicle lane change probability, and construct a vehicle multimodal lane change probability model;

[0027] S44. Based on constraints 1 and 2, construct an optimal matching model between the vehicle positioning data and the lanes of the road section where the bridge is located based on the matching priority;

[0028] S45. Lane matching is performed on each vehicle positioning according to different priorities, and the optimal matching model is used to solve the vehicle positioning point matching results of different sampling frequencies, and the point set of vehicle trajectory correction is integrated;

[0029] S46. Lane matching is performed on the vehicle positioning data of each bridge line in each bridge group, and the matching results of the vehicle positioning data of each bridge line in the road section are obtained;

[0030] S5. Obtain the spatiotemporal distribution of vehicles on the bridge deck based on a lane-level road network simulation model of the bridge cluster;

[0031] S51. Construct a lane-level road network simulation model for bridge clusters based on bridge design parameters and a microscopic traffic simulation model.

[0032] S52. Using a positioning point optimization matching model based on matching priority, the positioning data of each vehicle on the road section where the bridge group is located is matched with the matching results of each lane of the GIS road network topology, the time and speed of each vehicle entering the bridge are calculated, and the position conflicts of the vehicles are corrected.

[0033] S53. Based on the matching results, the bridge vehicle travel path is extracted and the simulation model parameters in the bridge lane-level road network simulation model are set;

[0034] S54. Run the lane-level road network simulation model to obtain the spatiotemporal distribution of vehicles. Simulate each bridge in the bridge group separately, and integrate the spatiotemporal distribution of vehicles on all bridges in the bridge group to obtain the spatiotemporal distribution of vehicles on the bridge deck.

[0035] S6. Match the vehicle load of vehicles passing through the bridge with the spatiotemporal distribution of vehicles on the bridge deck using vehicle positioning data to obtain the spatiotemporal distribution of the load on each bridge in the bridge group, and integrate them to obtain the spatiotemporal distribution of the load on the bridge group.

[0036] Furthermore, in said S12, the vehicle positioning data preprocessing includes missing value processing, error data processing and chronological sorting processing, and the GIS topological network section information includes section name, section ID, section road grade, section lane number and section lane direction;

[0037] In said S13, a bridge group is selected according to the selected area, and the bridge group is represented as b. Among them, e is the bridge type, k is the number of bridges, and the bridge group is matched with the GIS topological network section information to obtain the road section that the bridge group matches. The set of road sections that the bridge group matches with the GIS topological network is represented as r. The hidden Markov model is used to match the vehicle positioning data with the information of each road section in the GIS topology network, and the road sections matched by the vehicle positioning data are obtained. The vehicle positioning data of vehicle j and the road network GIS topology match the road section set, that is, the vehicle trajectory of vehicle j is represented as re j ,re j ={r1, r2, ..., r m}, where m is the number of road sections the vehicle passes through, and the set of road sections where the vehicle positioning data matches the GIS topology network, i.e., the complete vehicle trajectory, is represented as re, re = {re 1 ,re j ,…,re p}, where p is the number of vehicles;

[0038] In S14, the road section set r of the bridge group matching the GIS topology network is matched with the road section set re of the vehicle positioning data matching the GIS topology network according to the road section ID, and the vehicle trajectory passing through the road section where each bridge type of bridge in the bridge group is located is obtained.

[0039] Furthermore, in said S21, the vehicle load detection point information set is represented as ld, ld={ld 1 ,ld 2 ,…,ld r′}, where r′ is the number of load detection points. For each vehicle trajectory, the complete vehicle trajectory re of each vehicle is divided into multiple path sub-segments according to the road segment ID of the road segment where the vehicle load detection point information set is located. The vehicle trajectory re of vehicle j is j The set of path sub-segments after division is represented as rs j , Where a is the number of sub-segments of the vehicle trajectory, The oth path sub-segment after the vehicle trajectory of vehicle j is divided, and the path sub-segment division result of all vehicle trajectories is integrated and expressed as rs, rs = {rs 1 ,rs j ,…,rs p}, obtain the starting vehicle load detection point ld corresponding to the path sub-segment of each vehicle trajectory start The end point vehicle load detection point ld corresponding to the end end ;

[0040] In S22, for the regional bridge group, the path sub-segments of each bridge in the bridge group are matched according to the road segment set r that matches the bridge group with the GIS topological network and the path sub-segment division result rs of all vehicle trajectories, that is, the number of path sub-segments matched by bridge row, and the vehicle trajectory re of vehicle j. j The path sub-segment set rs j and bridges The matching results, i′=1, 2, …, k, that is, bridge The set of vehicle load sections matched by each line is expressed as Where c is the number of times vehicle j passes through bridge type e within the selected time, that is, the number of path sub-segments matched by the bridge type.

[0041] Furthermore, in said S31, the starting vehicle load detection point ld is determined according to the license plate number. start and the terminal vehicle load detection point ld end The corresponding data are filtered respectively to obtain the vehicle load of vehicle j passing the starting vehicle load detection point The detection time when vehicle j passes the starting vehicle load detection point Vehicle load of vehicle j passing the terminal vehicle load detection point and the detection time when vehicle j passes the terminal vehicle load detection point Where j′ is the number of times vehicle j passes the vehicle load detection point;

[0042] In said S32, the starting vehicle load detection point ld is selected start Or the end vehicle load detection point ldend It is defined as a specific vehicle load detection point. The projection coordinates of the specific vehicle load detection point are expressed as (x ld ,y ld ) The specific vehicle load detection point is u. The coordinates of the first vehicle positioning point before the specific vehicle load detection point are matched with the GIS topology network as (x u ,y u ), the time when the first vehicle location point data before a specific vehicle load detection point matches the GIS topology network is expressed as tp u The speed of the first vehicle location point data matching the GIS topology network before the specific vehicle load detection point is expressed as v u , the coordinates of the first vehicle location point data matching the GIS topology network after the specific vehicle load detection point are expressed as (x u+1 ,y u+1 ), the time when the first vehicle location point data after a specific vehicle load detection point matches the GIS topology network is expressed as tp u+1 The speed of the first vehicle location point data matching the GIS topology network after the specific vehicle load detection point is expressed as v u+1 , calculate the predicted time for the vehicle to pass a specific vehicle load detection point;

[0043] Predicted time t to pass a specific vehicle load detection point ld Expressed as:

[0044] Based on the predicted time t of passing a specific vehicle load detection point ld Get the predicted time when the vehicle passes the starting vehicle load detection point and the predicted time for the vehicle to pass the terminal vehicle load detection point

[0045] In the above S33, the predicted time t for passing the specific vehicle load detection point is set. ld The permissible error range from the actual time of detection at a specific vehicle load detection point is δ0;

[0046] When the error δ between the predicted time of passing the vehicle load detection point and the actual time of detection at a specific vehicle load detection point is less than the error allowable range δ0, it is expressed as:

[0047] Where ff is the minimum average error between the predicted time of the starting vehicle load detection point and the predicted time of the end vehicle load detection point, and h is the number of times the vehicle passes through the load section;

[0048] Obtain the vehicle load and time of the vehicle load detection point corresponding to all load sections within the allowable error range, and record them as the vehicle load of the actual starting vehicle load detection point Vehicle load at the actual endpoint vehicle load detection point Detection time of the actual starting point vehicle load detection point and actual endpoint

[0049] According to the actual starting point vehicle load test point vehicle load and the vehicle load at the actual endpoint vehicle load detection point Get the actual vehicle load passing through the bridge;

[0050] Actual vehicle load Expressed as:

[0051] Furthermore, in said S41, the positioning data of vehicles passing through the road section of each bridge type in the bridge group is obtained, and according to the number of lanes, lane length and GIS data of the road section of each bridge type, a GIS road network topology is constructed with the entrance of the road section of the bridge type as the starting point and the exit of the road section of the bridge type as the ending point;

[0052] In the above S42, for the vehicle positioning data of the road section where a single bridge row in the bridge group is located, the vehicle positioning data are sorted from high to low according to the sampling frequency, and a vehicle positioning sequence set S with different matching priorities is constructed, where S = {S1, S2, ..., S r″}, r″ is the number of vehicles in a single bridge direction passing through the bridge;

[0053] The Gaussian distribution function is used to evaluate the possibility of the vehicle positioning point in each lane. The lane set of the bridge section is bl, bl = [bl1, bl2, ..., bl s′}, where s′ is the number of lanes on the road section where the bridge is located. The shortest distance from the vehicle positioning point to the GIS road network topology of each lane and the possibility score of the vehicle positioning point in each lane are calculated respectively during driving in the same lane;

[0054] The probability score of vehicle positioning point k′ in lane i Expressed as:

[0055] Among them, δ 2 are Gaussian model parameters, which are obtained using moment estimation parameter estimation method based on historical data, i = 1, 2, …, s′;

[0056] The shortest distance between the vehicle positioning point k′ and the GIS road network topology corresponding to lane i Expressed as:

[0057] in, is the projection coordinate of the point on the GIS road network topology corresponding to lane i that is the shortest distance from the vehicle positioning point k′, (x k′ ,y k′ ) is the projection coordinate of the vehicle positioning point k′;

[0058] In S43, a Gaussian distribution model is used for fitting to identify the possibility of the vehicle changing lanes, based on the characteristic that the greater the difference in angle between the vehicle trajectory formed by the vehicle positioning point and the lane line shape, the smaller the possibility of the vehicle changing lanes;

[0059] The probability of not changing lanes when the vehicle positioning point k′ is in lane i Expressed as:

[0060] The angle between the vector formed by the current vehicle positioning point k′ and the previous vehicle positioning point and the linear shape of lane i Expressed as:

[0061] Among them, (x k-1 ,y k-1 ) is the projection coordinate of the previous point of the vehicle positioning point k′, (x k * ,y k * ) is the projection coordinate of the vehicle positioning point k′ (x k ,y k ) is the point with the shortest distance from the vehicle positioning point k′ on the current lane GIS road network topology, (x k-1 * ,y k-1 * ) is the projection coordinate of the previous point of the vehicle positioning point k′ (x k-1 ,y k-1 ) is the point with the shortest distance from the vehicle positioning point k′ on the current lane GIS road network topology, λ 2 For model parameters, the moment estimation method is used to estimate the parameters based on historical data;

[0062] The probability of a vehicle changing lanes is calculated based on the fact that the greater the distance between different lanes, the less likely the vehicle is to change lanes;

[0063] The probability of a vehicle changing from lane i to lane v Expressed as:

[0064] Among them, d i，v is the distance between lane i and lane v;

[0065] Integrate the multimodal lane-changing probability model of each vehicle positioning point in the vehicle positioning data

[0066] Among them, when v=i, the vehicle does not change lanes, and when v≠i, the vehicle changes lanes;

[0067] In S44, each vehicle positioning data point is matched with a lane, with the goal of maximizing the sum of the probability scores of each lane positioning point in the lane and the vehicle lane change probability during vehicle driving, thereby establishing a global optimal matching model between the vehicle positioning data and the lanes of the road section where the bridge is located;

[0068] The optimal matching model f between the vehicle positioning data and the lanes of the road section where the bridge is located is expressed as:

[0069] in, is the probability that the vehicle is in lane i, is the probability of a vehicle changing from lane i to lane v, N is the number of vehicle positioning points on the road section where the bridge is located, and q is the vehicle positioning point;

[0070] Constraint 1 is the lane change constraint. During the matching process, the lane change of the vehicle is constrained. The lane change direction of each lane of the bridge is judged to see whether it meets the number of lanes of the bridge. The lane bl is obtained. i The lane h(bl) where the vehicle is located after changing lanes using the lane changing direction a′ i , a′), if the corresponding lane does not exist on the bridge after the lane change direction a′ is used, it is expressed as 0;

[0071] The lane-changing constraint can be expressed as: h(bl i , a′)≠0;

[0072] Constraint 2 is the vehicle position constraint, which constrains the conflict of vehicle positions at the same time. The matching of each vehicle positioning data on the road section where the bridge is located is performed according to the sampling frequency of each vehicle positioning data as the priority. At the same time, the matching position of the unmatched vehicle positioning point does not conflict with the position of the previously matched vehicle. That is, the error between the matching position of the vehicle positioning point combined with the lane length occupied by the vehicle length and the position of the previously matched vehicle at the same time combined with the lane length occupied by the vehicle length should be less than the set error value. The coordinates of the vehicle positioning point k' to be matched are (bx k′ , by k′ ), the point with the shortest distance from the vehicle positioning point k′ to the GIS road network topology of lane i is The vehicle positioning point k′ corresponds to a vehicle length vl k′, the coordinates of the matched vehicle positioning point q at the same time as the vehicle positioning point k′ are (hx q ,hy q ), the point with the shortest distance from the vehicle positioning point q to the GIS road network topology of lane i is The vehicle length is vl q , the position conflict allowable error is χ;

[0073] The vehicle position constraint is expressed as:

[0074] In said S45, according to the constructed vehicle positioning sequence set S with different matching priorities, r′ , solve the optimal matching model in the order of the sequence, obtain the vehicle positioning point matching results of different sampling frequencies, including the matched lane number and the coordinate point closest to the vehicle positioning point on the GIS road network topology of the matched lane, and integrate the coordinate point closest to the vehicle positioning point on the GIS road network topology of the matched lane into the point set S″, S″={S1″,S2″,…,S r″ ″}.

[0075] Furthermore, in S51, for a bridge group, a lane-level traffic simulation network is established for each bridge row. For a single bridge, the bridge length, entrance, exit, lane width, and lane length are obtained based on the bridge design parameters, and a lane-level traffic simulation network model of the bridge deck is established with the bridge entrance as the starting point and the bridge exit as the end point.

[0076] In the above S52, the matching results of the positioning data of adjacent vehicles before and after each vehicle enters the bridge are intercepted and sorted according to the sampling frequency to determine the priority. According to the position of each lane at the bridge entrance, the data of two adjacent positioning points before and after each vehicle enters the bridge entrance and the position of the matching results on the lane are obtained from the matching results of each positioning point and each lane, and sorted according to the sampling frequency to obtain the matching result set si of the adjacent vehicle positioning points before and after each vehicle enters the bridge, si = {sv1, sv2, ..., sv q′}, q′ is the number of vehicles at the selected time;

[0077] A priority-based positioning point optimization matching model is used to obtain the matching results of each vehicle's positioning data on the road section where the bridge group is located and each lane of the GIS road network topology. The matching result sets of adjacent vehicle positioning points before and after each vehicle enters the bridge are processed in sequence, and the corresponding vehicle model and length are obtained based on the vehicle positioning data.

[0078] For the bridge alignment, the GIS road network topology data of each lane alignment of the bridge section is divided into multiple discrete points according to the set spatiotemporal sampling frequency, and the GIS road network topology data point set Gbx of each lane alignment of the bridge section is constructed. i , l is the number of GIS road network topology data points for lane i;

[0079] According to the matching result set si of the adjacent vehicle positioning points before and after each vehicle enters the bridge, the bridge entrance vehicle generation information is calculated in descending order of sampling frequency, that is, the time, speed and lane of a single vehicle entering the bridge. The coordinates of the adjacent vehicle positioning points of a single vehicle before the bridge entrance are (bx k′ ′,by k′ ′), the coordinates of the adjacent vehicle positioning points of a single vehicle after the bridge entrance are (bx k′+1 ′,by k′+1 ′), the coordinates of the adjacent vehicle positioning points of a single vehicle in front of the bridge entrance corresponding to the matched lane GIS road network topology are: The coordinates of the adjacent vehicle positioning points of a single vehicle after the bridge entrance corresponding to the matched lane GIS road network topology are: The detection time of a single vehicle at the adjacent vehicle positioning points before the bridge entrance is t u′ , the detection time of a single vehicle at adjacent vehicle positioning points after the bridge entrance is t u′+c , c=1,2,...n, the speed of a single vehicle at the adjacent vehicle positioning points before the bridge entrance is The vehicle speed of a single vehicle at the adjacent vehicle positioning points after the bridge entrance is The matching result point on the GIS road network topology of the lane at the bridge entrance is or b<c;

[0080] When the vehicle positioning point matching result and When in the same lane, i.e., when i = v, calculate the time when the vehicle enters the bridge;

[0081] The time t when the vehicle enters the bridge u′+b Expressed as:

[0082] t u′+b =t u′ +(t u′+c -t u′ )*

[0083] When the vehicle positioning point matching result and When the lanes are different, that is, when i≠v, the lane that matches the vehicle positioning point with the shortest distance to the bridge entrance is used as the lane for the vehicle to enter the bridge. The vehicle positioning point with the shortest distance to the bridge entrance is When , the time when the vehicle enters the bridge is t u′+b ;

[0084] The vehicle positioning point with the shortest distance to the bridge entrance is When , the time when the vehicle enters the bridge is t u′+b ;

[0085] The time t when the vehicle enters the bridge u′+b ′ is expressed as: t u′+b ′=t u′ +(t u′+c -t u′ )*

[0086] According to the vehicle positioning point Vehicle speed at and the vehicle positioning point is Vehicle speed at Get the average speed V′ of the vehicle when entering the bridge;

[0087] The average speed V′ of the vehicle when entering the bridge is expressed as:

[0088] Correct the time and lane of the bridge when the vehicle enters. When the time and lane of two vehicles entering the bridge conflict, the vehicle positioning point and vehicle positioning points In the same lane, when i=v, the vehicle speed is corrected by the correction parameter d, and the time when the vehicle enters the bridge is corrected until there is no lane conflict. The corrected speed and the corrected time when the vehicle enters the bridge are obtained.

[0089] The corrected speed V″ is expressed as:

[0090] The time when the vehicle enters the bridge is corrected as:

[0091] Vehicle positioning point and vehicle positioning points In different lanes, that is, when i≠v, the vehicle position conflict is resolved by modifying the lane the current vehicle is in. If there is a time or lane conflict after the lane modification, the vehicle position conflict is resolved by correcting the time to enter the bridge;

[0092] In S53, based on the optimized matching model of the vehicle positioning data with the matching priority and the lanes of the road section where the bridge is located, the matching results of the vehicle positioning data and each lane in the road section where the bridge is located in a single direction are obtained. According to the position of the bridge on the road section, the driving trajectory of the vehicle on the bridge deck in each lane is obtained and used as the vehicle driving path input in the lane-level road network simulation model. The model parameters include the simulation step size, the vehicle following model, and the vehicle lane changing model.

[0093] In the above S54, the bridge entrance vehicle generation information and the vehicle trajectory of the vehicle in each lane are input into the bridge lane-level road network simulation model, the simulation is run and the lane and longitudinal position of each vehicle on the bridge at different times are output, that is, the spatiotemporal distribution of each vehicle on the bridge deck, which is integrated to obtain the spatiotemporal distribution of vehicles on the bridge deck.

[0094] Furthermore, the simulation step size is obtained according to the time interval of the spatiotemporal distribution of vehicles on the bridge deck, the vehicle following model is a Wiedemann following model, and the lane changing model is a rule-based model.

[0095] The beneficial effects of the present application are as follows: the present application takes into account the vehicle correlation and time correlation between vehicle positioning data and vehicle load detection data, and combines the characteristic that the vehicle load remains unchanged within a certain period of time, and proposes a bridge vehicle load identification method that integrates vehicle positioning data and load detection points. The vehicle load detection points are used as nodes to divide the vehicle trajectory into multiple load sections, and the load section where the bridge is located is obtained. At the same time, according to the license plate and time characteristics, the vehicle positioning data in the load section where the bridge is located is correlated and matched with the vehicle load detection point data to obtain the actual load data of the vehicle when passing the bridge; the present application Please consider the advantages of all-weather and continuous trajectory of vehicle positioning data, integrate vehicle load data according to features such as license plates and spatiotemporal correlation, and realize the spatiotemporal distribution identification of loads on large-scale bridge groups in the city. This application makes full use of existing systems such as dynamic weighing systems and overweight control point systems, without the need to install dynamic weighing systems, video equipment and structural health monitoring systems, which greatly reduces monitoring costs and time consumption, and does not require manual operation; this application takes into account the fact that the higher the sampling frequency, the more accurate the feature extraction, and adopts a road lane target optimization matching method based on matching priority, with the sampling frequency as the basis for different vehicle positioning. The priority of data matching is to maximize the sum of the product of the probability of the vehicle positioning point being in the lane and the probability of the vehicle changing lanes during the process of the vehicle moving from the road section entrance to the road section exit. According to the vehicle lane changing constraints and vehicle position constraints, the optimal matching result between the vehicle positioning point and the road section lane is obtained, which provides accurate vehicle trajectories and the lanes where the bridge entrance is located for the spatiotemporal distribution identification method of bridge deck vehicles based on the microscopic traffic simulation model, thereby reducing the later identification error. This application can fully integrate the vehicle load detection data from different sources such as the overload control point system, the dynamic weighing system and the source overload control system. And vehicle positioning data of different vehicle types, including truck vehicle positioning data, taxi vehicle positioning data, online car-hailing vehicle positioning data, bus vehicle positioning data, two passenger and one dangerous goods vehicle positioning data, through the reuse and fusion of multi-source vehicle positioning data and multi-source vehicle load data, the data value is enhanced, and the spatiotemporal distribution of bridge surface vehicles can be identified; this application matches the vehicle load of vehicles passing through the bridge with the spatiotemporal distribution of bridge surface vehicles through vehicle positioning data, obtains the spatiotemporal distribution of load of each bridge in the bridge group, and integrates the spatiotemporal distribution of load of the bridge group, thereby realizing the identification of real-time vehicle load of the bridge group. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0097] Figure 1 is a flow chart of a method for identifying spatiotemporal load distribution of bridge groups based on multi-source data fusion. DETAILED DESCRIPTION

[0098] In order to make the technical solutions and advantages of the embodiments of the present application more clearly understood, the exemplary embodiments of the present application are further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, and are not an exhaustive list of all the embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other unless they conflict.

[0099] Referring to FIG1 , this embodiment is described in detail. A method for identifying spatiotemporal load distribution of a bridge group based on multi-source data fusion specifically includes the following steps:

[0100] S1. Perform GIS topology network matching on vehicles based on vehicle positioning data to obtain vehicle trajectory data;

[0101] S11. Select vehicle load detection points within the area and extract and collect vehicle load detection data and vehicle positioning data;

[0102] Specifically: Vehicle load detection points include overload control points, source overload control detection points, high-speed weight toll detection points and bridge dynamic weighing detection points. Vehicle load detection data and vehicle positioning data within the same selected time period are selected in the selected area. Vehicle load detection data, that is, the collected vehicle load detection point information includes license plate number, detection time, vehicle load, and the section ID, longitude and latitude of the section where the vehicle load is located. Vehicle positioning data includes vehicle length, vehicle model, vehicle type, license plate number, time, driving speed and latitude and longitude coordinates.

[0103] S12 pre-processes the vehicle positioning data and collects GIS topological network section information of the road network;

[0104] S13. Collect bridge group information and match vehicle positioning data with GIS topological network section information;

[0105] S14. According to the GIS topology network section information, obtain the vehicle positioning data matching each bridge line section in the bridge group to form vehicle trajectory data;

[0106] S2. Segment the vehicle trajectory based on the vehicle load detection point, and obtain the path section where the bridge is located in combination with the road section where the bridge is located;

[0107] S21. Divide all vehicle trajectories into multiple path sub-segments according to the road section ID of the vehicle load detection point;

[0108] S22. According to the bridge group information, the bridge line is located in the road section, extracting the bridge group of each bridge each bridge line where the path sub-section;

[0109] S3. Based on the vehicle positioning data and vehicle load data, identify the vehicle loads matching the start and end points of each line path segment of the bridge in the bridge group and calculate the actual vehicle load passing through the bridge;

[0110] S 31. For each vehicle load segment set matched to each bridge row, identify the starting and ending vehicle load detection points corresponding to the vehicle entry and exit segments of each path sub-segment, and calculate the detection time and vehicle load passing through the above two points;

[0111] S 32. Select a specific vehicle load detection point and calculate the predicted time of the specific vehicle load detection point, the predicted time of the vehicle passing the starting vehicle load detection point, and the predicted time of the vehicle passing the ending vehicle load detection point based on the vehicle positioning data;

[0112] S33. The vehicle positioning data is matched with the vehicle load data of the section from the starting point to the end point of the vehicle load detection point, ie, the load section, to obtain the actual vehicle load;

[0113] S4. Based on the vehicle positioning data and the vehicle load detection point detection data, identify the bridge line path section starting and ending points matching the vehicle load, and obtain the vehicle load passing the bridge;

[0114] S41. Select the vehicle positioning data of the road section where the bridge is located, and construct a lane-level GIS topology network of the road section where the bridge is located;

[0115] S42. Determine the matching priority based on the sampling frequency, construct a set of vehicle positioning sequences with different matching priorities for the vehicle positioning data of the road section where the bridge is located, and calculate the probability score of the vehicle positioning data point matching each lane;

[0116] S43. Use a Gaussian distribution model for fitting, identify the possibility of vehicle lane change, calculate the vehicle lane change probability, and construct a vehicle multimodal lane change probability model;

[0117] S44. Based on constraints 1 and 2, construct an optimal matching model between the vehicle positioning data and the lanes of the road section where the bridge is located based on the matching priority;

[0118] S45. Lane matching is performed on each vehicle positioning according to different priorities, and the optimal matching model is used to solve the vehicle positioning point matching results of different sampling frequencies, and the point set of vehicle trajectory correction is integrated;

[0119] S46. Lane matching is performed on the vehicle positioning data of each bridge line in each bridge group, and the matching results of the vehicle positioning data of each bridge line in the road section are obtained;

[0120] S5. Obtain the spatiotemporal distribution of vehicles on the bridge deck based on a lane-level road network simulation model of the bridge cluster;

[0121] S51. Construct a lane-level road network simulation model for bridge clusters based on bridge design parameters and a microscopic traffic simulation model.

[0122] S52. Using a positioning point optimization matching model based on matching priority, the positioning data of each vehicle on the road section where the bridge group is located is matched with the matching results of each lane of the GIS road network topology, the time and speed of each vehicle entering the bridge are calculated, and the position conflicts of the vehicles are corrected.

[0123] S53. Based on the matching results, the bridge vehicle travel path is extracted and the simulation model parameters in the bridge lane-level road network simulation model are set;

[0124] S54. Run the lane-level road network simulation model to obtain the spatiotemporal distribution of vehicles. Simulate each bridge in the bridge group separately, and integrate the spatiotemporal distribution of vehicles on all bridges in the bridge group to obtain the spatiotemporal distribution of vehicles on the bridge deck.

[0125] S6. Match the vehicle load of vehicles passing through the bridge with the spatiotemporal distribution of vehicles on the bridge deck using vehicle positioning data to obtain the spatiotemporal distribution of the load on each bridge in the bridge group, and integrate them to obtain the spatiotemporal distribution of the load on the bridge group.

[0126] Furthermore, in said S12, the vehicle positioning data preprocessing includes missing value processing, error data processing and chronological sorting processing, and the GIS topological network section information includes section name, section ID, section road grade, section lane number and section lane direction;

[0127] In said S13, a bridge group is selected according to the selected area, and the bridge group is represented as b. Where e is the bridge type, k is the number of bridges, and the bridge group is matched with the GIS topological network section information to obtain the road sections that match the bridge group. The set of road sections that match the bridge group and the GIS topological network is represented as r. The hidden Markov model is used to match the vehicle positioning data with the information of each road section in the GIS topology network, and the road sections matched by the vehicle positioning data are obtained. The vehicle positioning data of vehicle j and the road network GIS topology match the road section set, that is, the vehicle trajectory of vehicle j is represented as re j ,re j ={r1, r2, ..., r m}, where m is the number of road sections the vehicle passes through, and the set of road sections where the vehicle positioning data matches the GIS topology network, i.e., the complete vehicle trajectory, is represented as re, re = {re 1 ,re j ,…,re p}, where p is the number of vehicles;

[0128] In S14, the road section set r of the bridge group matched with the GIS topological network is matched with the road section set re of the vehicle positioning data matched with the GIS topological network according to the road section ID, and the vehicle trajectory passing through the road section where each bridge type of bridge in the bridge group is located is obtained;

[0129] Specifically, when the bridge has only one type of traffic, e is 1; when the bridge has two types of traffic, e is 1 or 2. The vehicle positioning data is obtained by collecting positioning data of vehicles such as trucks, buses, and two-passenger and one-hazardous vehicle, as well as vehicle navigation data.

[0130] Furthermore, in said S21, the vehicle load detection point information set is represented as ld, ld={ld 1 ,ld 2 ,…,ld r′}, where r′ is the number of load detection points. For each vehicle trajectory, the complete vehicle trajectory re of each vehicle is divided into multiple path sub-segments according to the road segment ID of the road segment where the vehicle load detection point information set is located. The vehicle trajectory re of vehicle j is j The set of path sub-segments after division is represented as rs j , Where a is the number of sub-segments of the vehicle trajectory, The oth path sub-segment after the vehicle trajectory of vehicle j is divided, and the path sub-segment division result of all vehicle trajectories is integrated and expressed as rs, rs = {rs 1 ,rs j ,…,rs p}, obtain the starting vehicle load detection point ld corresponding to the path sub-segment of each vehicle trajectory start The end point vehicle load detection point ld corresponding to the end end ;

[0131] In S22, for the regional bridge group, the path sub-segments of each bridge in the bridge group are matched according to the road segment set r that matches the bridge group with the GIS topological network and the path sub-segment division result rs of all vehicle trajectories, that is, the number of path sub-segments matched by bridge row, and the vehicle trajectory re of vehicle j. j The path sub-segment set rs ` and bridges The set of vehicle load sections matched by each line is expressed as Where c is the number of times vehicle j passes through bridge type e within the selected time, that is, the number of path sub-segments matched by the bridge type.

[0132] Furthermore, in said S31, the starting vehicle load detection point ld is determined according to the license plate number. start and the terminal vehicle load detection point ld end The corresponding data are filtered respectively to obtain the vehicle load of vehicle j passing the starting vehicle load detection point The detection time when vehicle j passes the starting vehicle load detection point Vehicle load of vehicle j passing the terminal vehicle load detection point and the detection time when vehicle j passes the terminal vehicle load detection point Where j′ is the number of times vehicle j passes the vehicle load detection point;

[0133] In said S32, the starting vehicle load detection point ld is selected start Or the end vehicle load detection point ld end It is defined as a specific vehicle load detection point. The projection coordinates of the specific vehicle load detection point are expressed as (x ld ,y ld ) The specific vehicle load detection point is u, and the coordinates of the first vehicle positioning point data matching the GIS topology network before the specific vehicle load detection point are expressed as (x u ,y u ), the time when the first vehicle location point data before a specific vehicle load detection point matches the GIS topology network is expressed as tp u The speed of the first vehicle location point data matching the GIS topology network before the specific vehicle load detection point is expressed as v u , the coordinates of the first vehicle location point data matching the GIS topology network after the specific vehicle load detection point are expressed as (x u+1 ,y u+1 ), the time when the first vehicle location point data after a specific vehicle load detection point matches the GIS topology network is expressed as tp u+1 The speed of the first vehicle location point data matching the GIS topology network after the specific vehicle load detection point is expressed as v u+1 , calculate the predicted time for the vehicle to pass a specific vehicle load detection point;

[0134] Predicted time t to pass a specific vehicle load detection point ld Expressed as:

[0135] Based on the predicted time t of passing a specific vehicle load detection point ld Get the predicted time when the vehicle passes the starting vehicle load detection point and the predicted time for the vehicle to pass the terminal vehicle load detection point

[0136] In the above S33, the predicted time t for passing the specific vehicle load detection point is set. ld The permissible error range from the actual time of detection at a specific vehicle load detection point is δ0;

[0137] When the error δ between the predicted time of passing the vehicle load detection point and the actual time of detection at a specific vehicle load detection point is less than the error allowable range δ0, it is expressed as:

[0138] Where ff is the minimum average error between the predicted time of the starting vehicle load detection point and the predicted time of the end vehicle load detection point, and h is the number of times the vehicle passes through the load section;

[0139] Obtain the vehicle load and time of the vehicle load detection point corresponding to all load sections within the allowable error range, and record them as the vehicle load of the actual starting vehicle load detection point Vehicle load at the actual endpoint vehicle load detection point Detection time of the actual starting point vehicle load detection point and the actual endpoint vehicle load detection point detection time

[0140] According to the actual starting point vehicle load test point vehicle load and the vehicle load at the actual endpoint vehicle load detection point Get the actual vehicle load passing through the bridge;

[0141] Actual vehicle load Expressed as:

[0142] Specifically, since a single vehicle may pass through the same vehicle load detection point multiple times, and the load size may also be different due to different times, the vehicle ID and detection time of the vehicle passing through the vehicle load detection point are combined to match the unique vehicle load when the vehicle passes through the vehicle load detection point, that is, the actual vehicle load.

[0143] Furthermore, in said S41, the positioning data of vehicles passing through the road section of each bridge type in the bridge group is obtained, and according to the number of lanes, lane length and GIS data of the road section of each bridge type, a GIS road network topology is constructed with the entrance of the road section of the bridge type as the starting point and the exit of the road section of the bridge type as the ending point;

[0144] In the above S42, for the vehicle positioning data of the road section where a single bridge row in the bridge group is located, the vehicle positioning data are sorted from high to low according to the sampling frequency, and a vehicle positioning sequence set S with different matching priorities is constructed, where S = {S1, S2, ..., S r″}, r″ is the number of vehicles in a single bridge direction passing through the bridge;

[0145] The Gaussian distribution function is used to evaluate the possibility of the vehicle positioning point in each lane. The lane set of the bridge section is bl, bl = {bl1, bl2, ..., bl s′}, where s′ is the number of lanes on the road section where the bridge is located. The shortest distance from the vehicle positioning point to the GIS road network topology of each lane and the possibility score of the vehicle positioning point in each lane are calculated respectively during driving in the same lane;

[0146] The probability score of vehicle positioning point k′ in lane i Expressed as:

[0147] Among them, δ 2 are Gaussian model parameters, which are obtained using moment estimation parameter estimation method based on historical data, i = 1, 2, …, s′;

[0148] The shortest distance between the vehicle positioning point k′ and the GIS road network topology corresponding to lane i Expressed as:

[0149] in, is the projection coordinate of the point on the GIS road network topology corresponding to lane i that is the shortest distance from the vehicle positioning point k′, (x k′ ,y k′ ) is the projection coordinate of the vehicle positioning point k′;

[0150] In S43, a Gaussian distribution model is used for fitting to identify the possibility of the vehicle changing lanes, based on the characteristic that the greater the difference in angle between the vehicle trajectory formed by the vehicle positioning point and the lane line shape, the smaller the possibility of the vehicle changing lanes;

[0151] The probability of not changing lanes when the vehicle positioning point k′ is in lane i Expressed as:

[0152] The angle between the vector formed by the current vehicle positioning point k′ and the previous vehicle positioning point and the linear shape of lane i Expressed as:

[0153] Among them, (x k-1 ,y k-1 ) is the projection coordinate of the previous point of the vehicle positioning point k′, (x k * ,y k * ) is the projection coordinate of the vehicle positioning point k′ (x k ,y k) is the point with the shortest distance from the vehicle positioning point k′ on the current lane GIS road network topology, (x k-1 * ,y k-1 * ) is the projection coordinate of the previous point of the vehicle positioning point k′ (x k-1 ,y k-1 ) is the point with the shortest distance from the vehicle positioning point k′ on the current lane GIS road network topology, λ 2 For model parameters, the moment estimation method is used to estimate the parameters based on historical data;

[0154] The probability of a vehicle changing lanes is calculated based on the fact that the greater the distance between different lanes, the less likely the vehicle is to change lanes;

[0155] The probability of a vehicle changing from lane i to lane v Expressed as:

[0156] Among them, d i，v is the distance between lane i and lane v;

[0157] Integrate the multimodal lane-changing probability model of each vehicle positioning point in the vehicle positioning data

[0158] Among them, when v=i, the vehicle does not change lanes, and when v≠i, the vehicle changes lanes;

[0159] In S44, each vehicle positioning data point is matched with a lane, with the goal of maximizing the sum of the probability scores of each lane positioning point in the lane and the vehicle lane change probability during vehicle driving, thereby establishing a global optimal matching model between the vehicle positioning data and the lanes of the road section where the bridge is located;

[0160] The optimal matching model f between the vehicle positioning data and the lanes of the road section where the bridge is located is expressed as:

[0161] in, is the probability that the vehicle is in lane i, is the probability of a vehicle changing from lane i to lane v, N is the number of vehicle positioning points on the road section where the bridge is located, and q is the vehicle positioning point;

[0162] Constraint 1 is the lane change constraint. During the matching process, the lane change of the vehicle is constrained. The lane change direction of each lane of the bridge is judged to see whether it meets the number of lanes of the bridge. The lane bl is obtained. i The lane h(bl) where the vehicle is located after changing lanes using the lane changing direction a′ i, a′), if the corresponding lane does not exist on the bridge after the lane change direction a′ is used, it is expressed as 0;

[0163] The lane-changing constraint can be expressed as: h(bl i , a′)≠0;

[0164] Constraint 2 is the vehicle position constraint, which constrains the conflict of vehicle positions at the same time. The matching of each vehicle positioning data on the road section where the bridge is located is performed according to the sampling frequency of each vehicle positioning data as the priority. At the same time, the matching position of the unmatched vehicle positioning point does not conflict with the position of the previously matched vehicle. That is, the error between the matching position of the vehicle positioning point combined with the lane length occupied by the vehicle length and the position of the previously matched vehicle at the same time combined with the lane length occupied by the vehicle length should be less than the set error value. The coordinates of the vehicle positioning point k' to be matched are (bx k′ , by k′ ), the point with the shortest distance from the vehicle positioning point k′ to the GIS road network topology of lane i is The vehicle positioning point k′ corresponds to a vehicle length vl k′ , the coordinates of the matched vehicle positioning point q at the same time as the vehicle positioning point k′ are (hx q ,hy q ), the point with the shortest distance from the vehicle positioning point q to the GIS road network topology of lane i is The vehicle length is vl q , the position conflict allowable error is χ;

[0165] The vehicle position constraint is expressed as:

[0166] In said S45, according to the constructed vehicle positioning sequence set S with different matching priorities, r′ , solve the optimal matching model in the order of the sequence, obtain the vehicle positioning point matching results of different sampling frequencies, including the matched lane number and the coordinate point closest to the vehicle positioning point on the GIS road network topology of the matched lane, and integrate the coordinate point closest to the vehicle positioning point on the GIS road network topology of the matched lane into the point set S″, S″={S1″,S2″,…,S r″ ″};

[0167] Specifically, after obtaining the vehicle positioning data of each bridge in the bridge group, it is necessary to obtain the vehicle's driving trajectory on different lanes of the bridge to provide vehicle trajectory data for the identification of the spatiotemporal distribution of bridge deck vehicle loads. By matching the vehicle positioning data of different sampling frequencies with the lane-level GIS topology data of the bridge section, the vehicle's driving trajectory on different lanes of the bridge is obtained. Due to the different sampling frequencies of different types of vehicle positioning data, the higher the frequency, the more accurate the features extracted from the vehicle positioning data, and the more accurate the lane change behavior identification and vehicle trajectory matching. Therefore, the sampling frequency is used as the judgment basis for the priority identification of vehicle positioning data. The higher the sampling frequency, the greater the priority of vehicle positioning data matching. At the same time, there may be conflicts in the matching positions of different vehicles at the same time. According to the matching priority of vehicle positioning data, conflicts in matching positions of different vehicles, and vehicle lane changes, a vehicle positioning optimization matching model based on matching priority is established to match vehicle positioning data of different sampling frequencies with the lanes of the bridge section, and obtain the lane corresponding to each positioning point and the position of the corresponding lane.

[0168] Furthermore, in S51, for a bridge group, a lane-level traffic simulation network is established for each bridge row. For a single bridge, the bridge length, entrance, exit, lane width, and lane length are obtained based on the bridge design parameters, and a lane-level traffic simulation network model of the bridge deck is established with the bridge entrance as the starting point and the bridge exit as the end point.

[0169] In the above S52, the matching results of the positioning data of adjacent vehicles before and after each vehicle enters the bridge are intercepted and sorted according to the sampling frequency to determine the priority. According to the position of each lane at the bridge entrance, the data of two adjacent positioning points before and after each vehicle enters the bridge entrance and the position of the matching results on the lane are obtained from the matching results of each positioning point and each lane, and sorted according to the sampling frequency to obtain the matching result set si of the adjacent vehicle positioning points before and after each vehicle enters the bridge, si = {sv1, sv2, ..., sv q′}, q′ is the number of vehicles at the selected time;

[0170] A priority-based positioning point optimization matching model is used to obtain the matching results of each vehicle's positioning data on the road section where the bridge group is located and each lane of the GIS road network topology. The matching result sets of adjacent vehicle positioning points before and after each vehicle enters the bridge are processed in sequence, and the corresponding vehicle model and length are obtained based on the vehicle positioning data.

[0171] For the bridge alignment, the GIS road network topology data of each lane alignment of the bridge section is divided into multiple discrete points according to the set spatiotemporal sampling frequency, and the GIS road network topology data point set Gbx of each lane alignment of the bridge section is constructed. i, l is the number of GIS road network topology data points for lane i;

[0172] According to the matching result set si of the adjacent vehicle positioning points before and after each vehicle enters the bridge, the bridge entrance vehicle generation information is calculated in descending order of sampling frequency, that is, the time, speed and lane of a single vehicle entering the bridge. The coordinates of the adjacent vehicle positioning points of a single vehicle before the bridge entrance are (bx k′ ′,by k′ ′), the coordinates of the adjacent vehicle positioning points of a single vehicle after the bridge entrance are (bx k′+1 ′,by k′+1 ′), the coordinates of the adjacent vehicle positioning points of a single vehicle in front of the bridge entrance corresponding to the matched lane GIS road network topology are: The coordinates of the adjacent vehicle positioning points of a single vehicle after the bridge entrance corresponding to the matched lane GIS road network topology are: The detection time of a single vehicle at the adjacent vehicle positioning points before the bridge entrance is t u′ , the detection time of a single vehicle at adjacent vehicle positioning points after the bridge entrance is t u′+c , c=1,2,...n, the speed of a single vehicle at the adjacent vehicle positioning points before the bridge entrance is The vehicle speed of a single vehicle at the adjacent vehicle positioning points after the bridge entrance is The matching result point on the GIS road network topology of the lane at the bridge entrance is or b<c;

[0173] When the vehicle positioning point matching result and When in the same lane, i.e., when i = v, calculate the time when the vehicle enters the bridge;

[0174] The time t when the vehicle enters the bridge u′+b Expressed as: t u′+b =t u′ +(t u′+c -t u′ )*

[0175] When the vehicle positioning point matching result and When the lanes are different, that is, when i≠v, the lane that matches the vehicle positioning point with the shortest distance to the bridge entrance is used as the lane for the vehicle to enter the bridge. The vehicle positioning point with the shortest distance to the bridge entrance is When , the time when the vehicle enters the bridge is t u′+b ;

[0176] The vehicle positioning point with the shortest distance to the bridge entrance is When , the time when the vehicle enters the bridge is t u′+b ';

[0177] The time t when the vehicle enters the bridge u′+b ′ is expressed as: t u′+b ′=t u′ +(t u′+c -t u′ )*

[0178] According to the vehicle positioning point Vehicle speed at and the vehicle positioning point is Vehicle speed at Get the average speed V′ of the vehicle when entering the bridge;

[0179] The average speed V′ of the vehicle when entering the bridge is expressed as:

[0180] Correct the time and lane of the bridge when the vehicle enters. When the time and lane of two vehicles entering the bridge conflict, the vehicle positioning point and vehicle positioning points In the same lane, when i=v, the vehicle speed is corrected by the correction parameter d, and the time when the vehicle enters the bridge is corrected until there is no lane conflict. The corrected speed and the corrected time when the vehicle enters the bridge are obtained.

[0181] The corrected speed V″ is expressed as:

[0182] The time when the vehicle enters the bridge is corrected as:

[0183] Vehicle positioning point and vehicle positioning points In different lanes, that is, when i≠v, the vehicle position conflict is resolved by modifying the lane the current vehicle is in. If there is a time or lane conflict after the lane modification, the vehicle position conflict is resolved by correcting the time to enter the bridge;

[0184] In S53, based on the optimized matching model of the vehicle positioning data with the matching priority and the lanes of the road section where the bridge is located, the matching results of the vehicle positioning data and each lane in the road section where the bridge is located in a single direction are obtained. According to the position of the bridge on the road section, the driving trajectory of the vehicle on the bridge deck in each lane is obtained and used as the vehicle driving path input in the lane-level road network simulation model. The model parameters include the simulation step size, the vehicle following model, and the vehicle lane changing model.

[0185] In S54, the bridge entrance vehicle generation information and the vehicle trajectory of the vehicle in each lane are input into the bridge lane-level road network simulation model, the simulation is run, and the lane and longitudinal position of each vehicle on the bridge at different times are output, that is, the spatiotemporal distribution of each vehicle on the bridge deck, which is integrated to obtain the spatiotemporal distribution of vehicles on the bridge deck;

[0186] Specifically, after obtaining the matching results of the vehicle positioning data of each bridge type in the bridge group, only the vehicle trajectory of each vehicle in the bridge type section can be obtained. Due to the different sampling frequencies of different vehicle positioning data, it is impossible to obtain the positions of all vehicles on the bridge deck in each lane at different times. The positioning data of each vehicle have inconsistent sampling time points and different numbers of vehicle positioning points for different vehicles. In order to obtain a spatiotemporal distribution of vehicles on the bridge deck that is close to the real one, a spatiotemporal distribution identification model of bridge deck vehicles is constructed based on a micro traffic simulation model; in the micro traffic simulation model, a single vehicle in the road network is taken as the research object, and the main research direction is the vehicle on the road. The impact of real micro-behaviors such as following, lane changing and overtaking between vehicles on the traffic capacity of the road network can dynamically simulate the real situation of vehicles performing different micro-behaviors under different road and traffic conditions. The main output of the micro-traffic simulation model is the instantaneous speed of each vehicle and the vehicle position in the road network. The micro-traffic simulation model mainly includes road network construction, vehicle generation module, signal control module, vehicle following module, and vehicle lane changing module. The bridge deck vehicle spatiotemporal distribution identification model improves the vehicle generation module of the micro-traffic simulation model based on vehicle positioning data with different sampling frequencies, and can obtain time and speed that are closer to the real vehicle entering the bridge.

[0187] Furthermore, the simulation step size is obtained according to the time interval of the spatiotemporal distribution of vehicles on the bridge deck, the vehicle following model is the Wiedemann following model, and the lane changing model is a rule-based model;

[0188] Specifically, the present application can select different time intervals for the spatiotemporal distribution of vehicles on the bridge deck according to actual needs. In this embodiment, the time interval for the spatiotemporal distribution of vehicles on the bridge deck is set to 0.2, in units of s, and the set spatiotemporal sampling frequency is to collect one point every 0.25 meters, dividing the bridge line into multiple points, and finding the GIS position of the bridge that matches the vehicle positioning point and the time of entering the bridge.

[0189] Although the present application has been described in terms of a limited number of embodiments, it will be apparent to those skilled in the art, having benefit of the foregoing description, that other embodiments are contemplated within the scope of the present application as described herein. Furthermore, it should be noted that the language used in this specification has been selected primarily for readability and didactic purposes, rather than for the purpose of explaining or limiting the subject matter of the present application. Consequently, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure made herein is illustrative and non-restrictive of the scope of the present application, and the scope of the present application is defined by the appended claims.

Claims

1. A method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion, characterized in that, Including the following steps: S1. Perform GIS topological network matching on the vehicle based on the vehicle positioning data to obtain vehicle trajectory data; S11. Select vehicle load detection points within a selected area, and extract and collect vehicle load detection data and vehicle positioning data; Specifically: The vehicle load detection points include overloading control points, source overloading control detection points, highway weighing toll detection points, and bridge dynamic weighing detection points. Select vehicle load detection data and vehicle positioning data within the same selected time in the selected area. The vehicle load detection data, that is, the information of the collected vehicle load detection points includes license plate number, detection time, vehicle load, and the section id, longitude, and latitude of the section where the vehicle load is located. The vehicle positioning data includes vehicle length, vehicle type, vehicle category, license plate number, time, driving speed, and longitude and latitude coordinates; S12. Preprocess the vehicle positioning data and collect the GIS topological network section information of the road network; S13. Collect bridge group information and match the vehicle positioning data with the GIS topological network section information; S14. According to the GIS topological network section information, obtain the vehicle positioning data matching the section of each bridge category in the bridge group to form vehicle trajectory data; S2. Segment the vehicle trajectory based on the vehicle load detection points, and combine the section where the bridge is located to obtain the path section where the bridge is located; S21. Divide all vehicle trajectories into multiple path sub-sections according to the section id of the section where the vehicle load detection point is located; S22. Extract the path sub-sections where each bridge of each bridge category in the bridge group is located according to the section where the bridge category in the bridge group information is located; S3. According to the vehicle positioning data and vehicle load data, identify the vehicle loads matching the start and end points of each bridge category path section in the bridge group, and calculate the actual vehicle load passing through the bridge; S31. For each set of vehicle load sections matching each bridge category, respectively identify the start vehicle load detection point and end vehicle load detection point corresponding to the vehicle entry section and exit section of each path sub-section, and calculate the detection time and vehicle load passing through the above two points; S32. Select a specific vehicle load detection point, and calculate the predicted time of the specific vehicle load detection point, the predicted time of passing through the start vehicle load detection point, and the predicted time of the vehicle passing through the end vehicle load detection point based on the vehicle positioning data; S33. Match the vehicle positioning data with the vehicle load data of the section from the start vehicle load detection point to the end vehicle load detection point, that is, the load section, to obtain the actual vehicle load; S4. According to the vehicle positioning data and the detection data of the vehicle load detection point, identify the vehicle loads matching the start and end points of the bridge category path section, and obtain the vehicle load passing through the bridge; S41. Select the vehicle positioning data of the section in the direction of the bridge category, and construct a lane-level GIS topological network of the section in the direction of the bridge category; S42. Determine the matching priority according to the sampling frequency, and construct vehicle positioning sequence sets with different matching priorities for the vehicle positioning data of the section where the bridge category is located, and calculate the possibility score of each vehicle positioning data point matching each lane; Construct sets of vehicle positioning sequences with different matching priorities for the vehicle positioning data of the section where the bridge category is located, and calculate the possibility score of each vehicle positioning data point matching each lane; S43. Fit using the Gaussian distribution model, identify the possibility of vehicle lane change, calculate the vehicle lane change probability, and construct a vehicle multi-modal lane change probability model; S44. According to Constraint 1 and Constraint 2, construct an optimal matching model for vehicle positioning data and the lanes of the section where the bridge is located based on the matching priority; S45. Perform lane matching on each vehicle positioning according to different priorities, use the optimal matching model to solve the matching results of vehicle positioning points with different sampling frequencies, and integrate to obtain the set of points for vehicle trajectory correction; S46. Perform lane matching on the vehicle positioning data of each bridge line type in the bridge group for the sections where the bridges are located, and obtain the matching results of the vehicle positioning data for the sections where each bridge line type is located; S5. Based on the lane-level road network simulation model of the bridge group, obtain the spatio-temporal distribution of vehicles on the bridge deck; S51. Based on the bridge design parameters and the microscopic traffic simulation model, construct a lane-level road network simulation model of the bridge group; S52. Use the optimal matching model of positioning points based on the matching priority to obtain the matching results of each vehicle positioning data and each lane of the GIS road network topology on the section where the bridge group is located, calculate the time and speed of a single vehicle entering the bridge, and correct the position conflict of the vehicle; S53. According to the matching results, extract the driving paths of vehicles on the bridge deck and set the simulation model parameters in the lane-level road network simulation model of the bridge; S54. Run the lane-level road network simulation model to obtain the spatio-temporal distribution of vehicles. Perform simulations on each bridge in the bridge group respectively, and integrate the spatio-temporal distribution of vehicles on all bridges in the bridge group to obtain the spatio-temporal distribution of vehicles on the bridge deck; S6. Match the vehicle loads of vehicles passing through the bridge with the spatio-temporal distribution of vehicles on the bridge deck through vehicle positioning data, obtain the spatio-temporal distribution of loads for each bridge in the bridge group, and integrate to obtain the spatio-temporal distribution of loads for the bridge group.

2. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 1, wherein In the above S12, the preprocessing of vehicle positioning data includes missing value processing, error data processing, and sorting processing in chronological order. The GIS topological network section information includes section name, section id, section road grade, number of lanes in the section, and lane direction; In S13, a bridge group is selected according to the selected area, and the bridge group is denoted as b. Among them, e is the bridge line type and k is the number of bridges. The bridge group is matched with the GIS topological network section information to obtain the sections matched by the bridge group. The set of sections where the bridge group is matched with the GIS topological network is denoted as r. The vehicle positioning data is matched with the information of each section in the GIS topological network by using the Hidden Markov Model to obtain the section where the vehicle positioning data is matched. The set of sections where the vehicle positioning data of vehicle j is matched with the road network GIS topology, that is, the vehicle trajectory of vehicle j is represented as re j , re j ={r1, r2, …, r m}, where m is the number of sections passed by the vehicle. The set of sections where the vehicle positioning data is matched with the GIS topological network, that is, the complete vehicle trajectory is represented as re, re = {re 1 , re j , …, re p}, where p is the number of vehicles; In the above S14, according to the section id, match the section set r where the bridge group is matched with the GIS topological network and the section set re where the vehicle positioning data is matched with the GIS topological network to obtain the vehicle trajectories passing through the sections where each bridge line type in the bridge group is located.

3. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 2, wherein In S21, the vehicle load detection point information set is denoted as ld, ld = {ld 1 , ld 2 , …, ld r′}, where r′ is the number of load detection points. For each vehicle trajectory, according to the section id of the section where the vehicle load detection point information set is located, the complete vehicle trajectory re of each vehicle is divided into multiple path sub - sections. The set of path sub - sections after division of the vehicle trajectory re j of vehicle j is denoted as rs j , where a is the number of path sub - section divisions of the vehicle trajectory, is the o - th path sub - section after division of the vehicle trajectory of vehicle j. The result of the division of the path sub - sections of all vehicle trajectories obtained by integration is denoted as rs, rs = {rs 1 , rs j , …, rs p}, and the starting point vehicle load detection point ld start corresponding to the start of the path sub - section of each vehicle trajectory and the ending point vehicle load detection point ld end corresponding to the end are obtained; In S22, for the regional bridge group, according to the set of road segments r that the bridge group matches with the GIS topological network and the result rs of the path sub-segment division of all vehicle trajectories, match the path sub-segments where each bridge in the bridge group is located, that is, the number of path sub-segments matched by the bridge type, and the path sub-segment set rs of the vehicle trajectory re of vehicle j j of the vehicle trajectory re j and the bridge The matching results, where i′ = 1, 2, …, k, i.e., the bridge The set of each vehicle load section matched by each row is expressed as Among them, c is the number of times vehicle j passes through bridge line type e within the selected time, that is, the number of path sub-sections matched by the bridge line type.

4. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 3, characterized in that, In S31, the data corresponding to the starting vehicle load detection point ld start and the ending vehicle load detection point ld end are screened respectively to obtain the vehicle load of vehicle j passing through the starting vehicle load detection point the detection time of vehicle j passing through the starting vehicle load detection point the vehicle load of vehicle j passing through the ending vehicle load detection point and the detection time of vehicle j passing through the ending vehicle load detection point where j′ is the number of times vehicle j passes through the vehicle load detection point; In S32, the selected starting vehicle load detection point ld start or the ending vehicle load detection point ld end is defined as a specific vehicle load detection point, and the projected coordinates of the specific vehicle load detection point are expressed as (x ld， y ld ). The specific vehicle load detection point is u, and the point coordinates of the first vehicle positioning point before the specific vehicle load detection point matching the GIS topological network are represented as (x u , y u ), the time of the first vehicle positioning point before the specific vehicle load detection point matching the GIS topological network is represented as tp u , the speed of the first vehicle positioning point before the specific vehicle load detection point matching the GIS topological network is represented as v u , the point coordinates of the first vehicle positioning point after the specific vehicle load detection point matching the GIS topological network are represented as (x u+1 , y u+1 ), the time of the first vehicle positioning point after the specific vehicle load detection point matching the GIS topological network is represented as tp u+1 and the speed of the first vehicle positioning point after the specific vehicle load detection point matching the GIS topological network is represented as v u+1 , and the predicted time for the vehicle to pass through the specific vehicle load detection point is calculated; Prediction time t for a specific vehicle load detection point ld Expressed as: Based on the predicted time t when passing through a specific vehicle load detection point ld Obtain the predicted time when the vehicle passes through the starting point vehicle load detection point and the predicted time when the vehicle passes the terminal vehicle load detection point In S33, a prediction time t passing through a specific vehicle load detection point is set. ld The allowable error range from the actual detection time of the specific vehicle load detection point is δ0. When the error δ between the predicted time when passing through the vehicle load detection point and the actual detection time of a specific vehicle load detection point is less than the allowable error range δ0, it is expressed as: Among them, ff is the minimum average error between the predicted time of the starting vehicle load detection point and the predicted time of the ending vehicle load detection point, and h is the number of times the vehicle passes through the load section; Obtain the vehicle load and time of the vehicle load detection points corresponding to all load sections within the allowable error range, and record it as the vehicle load of the actual starting point vehicle load detection point Vehicle load at the vehicle load detection point of the actual end point Detection time of the detection point of the actual starting vehicle load and the detection time of the actual end vehicle load detection point Vehicle load at the vehicle load detection point based on the actual starting point and the actual ending vehicle load detection Vehicle load at a point to obtain the actual vehicle load passing through the bridge; Actual vehicle load Expressed as:

5. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 4, characterized in that In the above S41, obtain the vehicle positioning data passing through the sections where each bridge line type in the bridge group is located. According to the number of lanes, lane length, and GIS data in the sections where each bridge line type is located, construct a GIS road network topology with the entrance of the section where the bridge line type is located as the starting point and the exit of the section where the bridge line type is located as the ending point; In S42, for the vehicle positioning data of the section where the single bridge type in the bridge group is located, the vehicle positioning data is sorted from high to low according to the sampling frequency, and a set S of vehicle positioning sequences with different matching priorities is constructed, S = {S1, S2, …, S r″}, where r″ is the number of vehicles passing through the direction of the single bridge type of the bridge; The Gaussian distribution function is used to evaluate the possibility of the vehicle positioning point in each lane. The lane set of the section where the bridge is located is bl, and bl = {bl1, bl2, …, bl s′}, where S′ is the number of lanes in the section where the bridge is located. Calculate the shortest distance from the vehicle positioning point to the GIS road network topology of each lane and the possibility score of the vehicle positioning point in each lane during the driving process in the same lane; Possibility score of vehicle positioning point k′ in lane i Expressed as: Among them, δ 2 is a Gaussian model parameter, which is obtained by using the moment estimation parameter estimation method based on historical data, where i = 1, 2, …, s′; The shortest distance between the vehicle positioning point k′ and the GIS road network topology corresponding to lane i Expressed as: Among them, is the projected coordinate of the point on the GIS road network topology corresponding to lane i that is at the shortest distance from the vehicle positioning point k′, and (x k′ , y k′ ) is the projected coordinate of the vehicle positioning point k′; In S43, according to the characteristic that the greater the difference between the vehicle trajectory formed by the vehicle positioning points and the lane alignment, the smaller the possibility of vehicle lane change, a Gaussian distribution model is used for fitting to identify the possibility of vehicle lane change; The probability of not changing lanes when the vehicle positioning point k′ is on lane i Expressed as: The angle between the vector formed by the current vehicle positioning point k′ and the previous vehicle positioning point and the alignment of lane i Expressed as: Among them, (x k-1 , y k-1 ) is the projection coordinates of the previous point of the vehicle positioning point k′, (x k * , y k * ) is the point on the current lane GIS road network topology that has the shortest distance from the projection coordinates (x k , y k ) of the vehicle positioning point k′ to the vehicle positioning point k′, (x k-1 * , y k-1 * ) is the point on the current lane GIS road network topology that has the shortest distance from the projection coordinates (x k-1 , y k-1 ) of the previous point of the vehicle positioning point k′ to the vehicle positioning point k′, λ 2 is a model parameter, which is obtained by parameter estimation using the method of moment estimation based on historical data; According to the fact that the greater the distance between different lanes, the smaller the possibility of vehicle lane change, calculate the vehicle lane change probability; Probability that a vehicle changes from lane i to lane v Expressed as: where d i,v is the distance between lane i and lane v; Integrate to obtain the multi-modal lane-changing probability model for each vehicle positioning point in the vehicle positioning data Among them, when v = i, the vehicle does not change lanes, and when v ≠ i, the vehicle changes lanes; In S44, each vehicle positioning data point is matched with the lane. With the goal of maximizing the sum of the product of the possibility scores of each lane positioning point of the vehicle during driving in the lane and the vehicle lane change probability, Establish a global optimization matching model between the vehicle positioning data and the lanes of the section where the bridge is located; The optimization matching model f between vehicle positioning data and lanes of the road section where the bridge is located is expressed as: Among them, The probability of the vehicle being in lane i, is the probability that the vehicle changes from lane i to lane v, N is the number of positioning points of the vehicle in the section where the bridge is located, and q is the vehicle positioning point; Constraint condition 1 is a lane change constraint, which restricts vehicle lane changes during the matching process. It determines whether each lane change direction of each lane of the bridge conforms to the number of lanes of the bridge, and obtains lane bl i The lane h(bl i , a′) where the vehicle is located after changing lanes using the lane change direction a′. If there is no corresponding lane on the bridge after changing lanes using the lane change direction a′, it is represented as 0; The lane change constraint is expressed as: h(bl i , a′) ≠ 0; Constraint condition 2 is the vehicle position constraint, which constrains the vehicle position conflicts at the same moment. The matching of the positioning data of each vehicle in the section where the bridge is located is prioritized according to the sampling frequency of each vehicle's positioning data. At the same moment, the matching position of the uncompleted vehicle positioning points does not conflict with the point positions of the vehicles that have been matched previously, that is, the error between the lane length occupied by the matching position of the vehicle positioning point combined with the vehicle length and the lane length occupied by the point positions of the vehicles that have been matched previously at the same moment should be less than the set error value. The coordinates of the vehicle positioning point k' to be matched are (bx k′ , by k′ ), and the point corresponding to the vehicle positioning point k' that is matched to the shortest distance of the GIS road network topology in lane i is The vehicle positioning point k' corresponds to a vehicle length of vl k′ , and the coordinates of the matched vehicle positioning point q at the same moment as the vehicle positioning point k' are (hx q , hy q ). The point corresponding to the vehicle positioning point q on the GIS road network topology with the shortest distance to the matched lane i is The vehicle length is vl q , and the allowable error for position conflict is χ; The vehicle position constraint is expressed as: In S45, according to the set S of vehicle positioning sequences constructed with different matching priorities r′ , the optimal matching model is solved in the order of the sequences to obtain the vehicle positioning point matching results at different sampling frequencies, including the matched lane numbers and the coordinate points on the lane GIS road network topology that are closest to the vehicle positioning points. The coordinate points on the lane GIS road network topology that are closest to the vehicle positioning points are integrated into the point set S″ for vehicle trajectory correction, S″ = {S1″, S2″, …, S r″ ″}.

6. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 5, wherein In S51, for a bridge group, a lane-level traffic simulation road network is established for each bridge type. For a single bridge, according to the bridge design parameters, obtain the bridge length, entrance, exit, the width of each lane, and the length of each lane. Taking the bridge entrance as the starting point and the bridge exit as the ending point, establish a lane-level traffic simulation road network model of the bridge deck; In S52, the adjacent vehicle positioning data matching results before and after each vehicle enters the bridge are intercepted, sorted according to the sampling frequency to determine the priority. According to the positions of each lane at the bridge entrance, two positioning point data adjacent before and after each vehicle enters the bridge entrance and the positions of the matching results on the lanes are obtained from the matching results of each positioning point and each lane, and sorted according to the sampling frequency to obtain the adjacent vehicle positioning point matching result set si of each vehicle before and after entering the bridge, si = {sv1, sv2,..., sv q′}, q ' is the number of vehicles at the selected time; Adopt an optimization matching model of positioning points based on matching priority to obtain the matching results between each vehicle positioning data and each lane of the GIS road network topology in the section where the bridge group is located. The matching result sets of adjacent vehicle positioning points before and after each vehicle enters and exits the bridge are processed sequentially, and the corresponding vehicle type and vehicle length are obtained according to the vehicle positioning data; For the alignment of bridge line types, according to the set spatio-temporal sampling frequency, the GIS road network topology data of the alignments of each lane in the section where the bridge is located are divided into multiple discrete points, and a set Gbx of GIS road network topology data points for the alignments of each lane in the section in the direction of the bridge line type is constructed. i , l is the number of GIS road network topology data points of lane i; According to the set si of adjacent vehicle positioning point matching results before and after each vehicle enters the bridge, in the order of sampling frequencies from high to low, the information generated by the vehicle at the bridge entrance is calculated, that is, the time, speed, and lane when a single vehicle enters the bridge. The coordinates of the adjacent vehicle positioning point of a single vehicle before the bridge entrance are (bx k′ ′, by k′ ′), and the coordinates of the adjacent vehicle positioning point of a single vehicle after the bridge entrance are (bx k′+1 ′, by k′+1 ′). The coordinates of the point on the lane GIS road network topology corresponding to the adjacent vehicle positioning point of a single vehicle before the bridge entrance are The point coordinates on the lane GIS road network topology corresponding to the adjacent vehicle positioning points after the single vehicle enters the bridge are The detection time of the adjacent vehicle positioning point before the bridge entrance for a single vehicle is t u′ , the detection time of the adjacent vehicle positioning point after the bridge entrance for a single vehicle is t u′+c , c = 1, 2,... n, the vehicle speed of the adjacent vehicle positioning point before the bridge entrance for a single vehicle is The vehicle speed of the adjacent vehicle positioning point after the single vehicle enters the bridge is The matching result points on the GIS road network topology of the lanes at the bridge entrance are or When the vehicle positioning point matching result And When in the same lane, that is, when i = v, calculate the time when the vehicle enters the bridge; The time t when the vehicle enters the bridge u′+b It is expressed as: When the vehicle positioning point matching result and When in different lanes, that is, when i≠v, the lane that matches the vehicle positioning point with the shortest distance to the bridge entrance is used as the lane for the vehicle to enter the bridge, and the vehicle positioning point with the shortest distance to the bridge entrance is When the time is t, the vehicle enters the bridge u′+b ; The vehicle positioning point with the shortest distance to the bridge entrance is When the vehicle enters the bridge, the time is t u′+b ′; The time t when the vehicle enters the bridge u′+b is expressed as: Based on the vehicle positioning point as Vehicle speed at and the vehicle positioning point is Vehicle speed at Obtain the average speed V' of the vehicle when it enters the bridge; The average speed V′ when the vehicle enters the bridge is expressed as: Correct the time and lane of the bridge when the vehicle enters. When there is a conflict in the time and lane when two vehicles enter the bridge, that is, the vehicle positioning point and vehicle positioning points When in the same lane, i = v, correct the vehicle speed by modifying the parameter d, and then correct the time when the vehicle enters the bridge until there is no lane conflict, and obtain the corrected speed and the corrected time when the vehicle enters the bridge; The corrected speed V″ is expressed as: The corrected time for the vehicle to enter the bridge is expressed as: Vehicle positioning point and vehicle positioning points When in different lanes, that is, i ≠ v, solve the vehicle position conflict by modifying the lane where the current vehicle is located. When there is a conflict in time or lane after modifying the lane, solve the vehicle position conflict by correcting the time when entering the bridge; In S53, based on the optimization matching model between the vehicle positioning data and the lanes of the section where the bridge is located with matching priority, obtain the matching results between the vehicle positioning data and each lane in the section where the single bridge type direction is located. According to the position of the bridge in this section, obtain the driving trajectory of the vehicle on each lane on the bridge deck, and use it as the driving path input of the vehicle in the lane-level road network simulation model. The model parameters include the simulation step size, the vehicle following model, and the vehicle lane change model; In S54, input the vehicle generation information at the bridge entrance and the vehicle trajectories of each vehicle in each lane into the bridge lane-level road network simulation model, run the simulation and output the lanes and longitudinal positions of each vehicle on the bridge at different times, that is, the spatio-temporal distribution of each vehicle on the bridge deck, and integrate to obtain the spatio-temporal distribution of the vehicles on the bridge deck.

7. The method for identifying the spatio-temporal load distribution of a bridge group based on multi-source data fusion according to claim 6, wherein The simulation step size is obtained according to the time interval of the spatio-temporal distribution of vehicles on the bridge deck. The vehicle following model is the Wiedemann following model, and the lane-changing model is a rule-based model.

Images