OD quantity estimation method, device and electronic equipment for regional trunk line network

Abstract

The present invention provides an OD amount estimation method, device and electronic device for a regional trunk road network, comprising: determining the flow type of the basic road section adjacent to each path node in the regional trunk road network according to a preset direction from a starting point to an end point; estimating the preliminary lane change probability of the path node for training a probability graph model according to a calculation method corresponding to the flow type of the basic road section adjacent to each path node in the regional trunk road network; inputting the intersection flow and the speed of the target object into the trained probability graph model, calculating the lane change probability of the path node and correcting it; calculating the OD amount of each path in the regional trunk road network according to the corrected lane change probability of each path node in the regional trunk road network; the embodiment of the present invention determines the corresponding lane change probability calculation method according to the traffic flow type of various trunk roads in the regional trunk road network, so as to alleviate the technical problem of low accuracy of OD amount estimation results in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of OD traffic volume estimation of a road network, and in particular to an OD volume estimation method, device and electronic equipment for a regional trunk road network. Background Art

[0002] Highway OD data can be understood as the traffic volume with known starting points and known destinations, reflecting the distribution of regional traffic demand, and playing an important role in road network planning and highway construction. Early OD matrix estimation required large-scale traffic surveys, which consumed a lot of resources, and the relevant data were difficult to obtain and inaccurate. With the development of big data, with the help of multi-source big data such as mobile phone signaling and vehicle card swiping, the basic distribution matrix of regional incomplete OD can be obtained, and the construction of high-speed toll stations and the popularization of traffic detectors can obtain relatively high-quality traffic detection data (location, speed, starting and ending points, etc.).

[0003] In the existing OD estimation research, expressways or trunk roads are often taken separately as the research objects, that is, only the OD distribution of roads of the same level is studied, while there are few studies on the OD distribution of regional trunk road networks composed of expressways and trunk roads. The inventors have found that there are currently methods for estimating the uncertainty of OD of urban road network traffic flow based on vehicle license plate recognition data, and methods for generating OD of provincial highway networks based on expressway toll data and traffic control station data. However, the accuracy of the OD results estimated by the above two methods is greatly affected by historical OD data or prior matrices. In practical applications, it is impossible to accurately require the complete reliability of historical OD data or prior matrices. Therefore, the accuracy of the OD results estimated by the above methods is general. Summary of the invention

[0004] The purpose of the present invention is to provide an OD quantity estimation method, device and electronic equipment for a regional trunk road network, and to determine the corresponding lane change probability calculation method by the traffic flow types of various trunk roads in the regional trunk road network, so as to alleviate the technical problem of low accuracy of OD quantity estimation results in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides an OD volume estimation method for a regional trunk road network, wherein the regional trunk road network is an area where multiple paths intersect between at least one starting point and at least one end point; the intersection of the paths is a path node; each path node is adjacent to at least one basic road segment; the driving flow of a target object on the same basic road segment is a cross-sectional flow; the lane change flow between different basic road segments of the target object is an intersection flow; the method comprises:

[0006] Determine the flow type of the basic road section adjacent to each path node in the regional trunk road network according to the preset direction from the starting point to the end point; wherein the flow type includes the intersection flow unknown and the intersection flow known;

[0007] According to the calculation method corresponding to the traffic type of the basic road section adjacent to each path node in the regional trunk road network, the preliminary lane change probability of the path node is estimated; wherein the preliminary lane change probability is used to train the probability graph model;

[0008] Inputting the intersection flow and the speed of the target object into a trained probability graph model to calculate the lane change probability of the path node;

[0009] According to the lane change probability of each path node in the regional trunk road network, the OD value of each path in the regional trunk road network is calculated.

[0010] In combination with the first aspect, an embodiment of the present invention provides a first possible implementation of the first aspect, wherein the step of estimating the preliminary lane change probability of the path node according to the calculation method corresponding to the flow type of the basic road section adjacent to each path node in the regional trunk road network includes:

[0011] If the intersection flow of the basic road section adjacent to the path node is known, then the preliminary lane change probability of the path node is calculated according to the proportional relationship between the intersection flow and the cross-sectional flow of each basic road section adjacent to the path node;

[0012] If the intersection flow of the basic road section adjacent to the path node is unknown, the preliminary lane changing probability of the path node is calculated according to the lane changing influencing factors of the path node and the logistic regression model.

[0013] In combination with the first aspect, an embodiment of the present invention provides a second possible implementation of the first aspect, wherein if the intersection flow of the basic road section adjacent to the path node is unknown, the step of estimating the preliminary lane change probability of the path node according to the lane change influencing factors of the path node and the logistic regression model includes:

[0014] Determining the utility of each basic road segment adjacent to the lane change of the path node according to the lane change influencing factors of the path node;

[0015] Based on the utility of each basic road segment adjacent to the path node change in a logistic regression model, a preliminary lane change probability of each basic road segment adjacent to the path node change is calculated.

[0016] In combination with the first aspect, an embodiment of the present invention provides a third possible implementation of the first aspect, wherein the step of inputting the intersection flow and the speed of the target object into a trained probabilistic graph model and calculating the lane change probability of the path node includes:

[0017] The intersection flow and the speed of the target object of the basic road section adjacent to the path node are input into the trained probability graph model, and the lane change probability of each basic road section adjacent to each path node in the regional trunk road network is calculated respectively; wherein the trained probability graph model includes the calibrated road type, the number of lanes, and the corresponding weights of the univariate local feature function and the binary transfer feature function.

[0018] In combination with the first aspect, an embodiment of the present invention provides a fourth possible implementation manner of the first aspect, wherein, before the step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network, the method further includes:

[0019] According to the preset direction from the starting point to the end point, if the observed road on which the target object has changed direction before the current path node includes a highway, the lane change probability of each basic road segment adjacent to the current path node is corrected by a preset exponential function.

[0020] In combination with the first aspect, an embodiment of the present invention provides a fifth possible implementation of the first aspect, wherein, before the step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network, the method further includes:

[0021] If there is a lane changing probability of the path node calculated according to the trained probability graph model, then the objective function of the estimated flow rate of each basic road segment adjacent to the path node is determined according to the lane changing probability of the path node calculated according to the trained probability graph model; the lane changing probability of the path node calculated according to the trained probability graph model is corrected so that the difference between the estimated flow rate of each basic road segment adjacent to the path node and the intersection flow rate of each basic road segment adjacent to the path node meets the target expectation.

[0022] In combination with the first aspect, an embodiment of the present invention provides a sixth possible implementation of the first aspect, wherein the lane change probability includes a corrected lane change probability and an uncorrected lane change probability; and the step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network includes:

[0023] The OD amount of the path from each starting point to each end point in the regional trunk road network is calculated according to the product of the flow corresponding to each starting point in the regional trunk road network and the lane change probability of each basic road section adjacent to each path node change.

[0024] In a second aspect, an embodiment of the present invention further provides an OD volume estimation device for a regional trunk road network, wherein the regional trunk road network is an area where multiple paths intersect between at least one starting point and at least one end point; the intersection of the paths is a path node; each path node is adjacent to at least one basic road section; the driving flow of the target object on the same basic road section is the cross-sectional flow; the lane change flow between different basic road sections of the target object is the intersection flow; the device comprises:

[0025] A determination module, which determines the flow type of the basic road section adjacent to each path node in the regional trunk road network according to a preset direction from the starting point to the end point; wherein the flow type includes the intersection flow type being unknown and the intersection flow type being known;

[0026] A first estimation module estimates a preliminary lane change probability of each path node according to a calculation method corresponding to a flow type of a basic road section adjacent to each path node in the regional trunk road network; wherein the preliminary lane change probability is used to train a probability graph model;

[0027] A second estimation module inputs the intersection flow and the speed of the target object into a trained probability graph model to calculate the lane change probability of the path node;

[0028] A calculation module calculates the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network.

[0029] In a third aspect, an embodiment provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the steps of the method described in any of the aforementioned implementation methods are implemented.

[0030] In a fourth aspect, an embodiment provides a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the steps of the method described in any one of the aforementioned implementations.

[0031] The embodiment of the present invention provides an OD amount estimation method, device and electronic device for a regional trunk road network. First, the preset direction corresponding to the starting point to the end point is determined according to the driving direction of the target object in the regional trunk road network, and then the traffic type of the basic section adjacent to each path node in the regional trunk road network is obtained according to the road combination scene of the basic section adjacent to each path node; secondly, the initial lane change probability calculation method of the corresponding path node can be selected according to the traffic type of the basic section adjacent to each path node. For example, if the intersection flow of the basic section adjacent to the path node is unknown, the initial lane change probability of the path node is calculated according to the logistic regression model. For example, if the intersection flow of the basic section adjacent to the path node is known, the initial lane change probability is directly calculated. Based on the calculated initial lane change probability of each path node, a probability graph model is trained, and the collected intersection flow and the driving speed of the target object are input into the trained probability graph model to calculate the lane change probability of the path node; by calculating the lane change probability of each path node in the regional trunk road network in this way, the OD amount of each path in the regional trunk road network can be estimated more accurately.

[0032] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A flow chart of a method for estimating the OD quantity of a regional trunk line network provided by an embodiment of the present invention;

[0036] Figure 2 A schematic diagram of the relationship between a basic road segment and a path node provided by an embodiment of the present invention;

[0037] Figure 3 A basic road section consisting of a highway provided in an embodiment of the present invention;

[0038] Figure 4A symmetrical basic road section composed of a mixture of expressways and national and provincial roads provided in an embodiment of the present invention;

[0039] Figure 5 Another symmetrical basic road section provided by an embodiment of the present invention is a mixture of expressways and national and provincial roads;

[0040] Figure 6 An asymmetric basic road section provided by an embodiment of the present invention is a mixture of expressways and national and provincial roads;

[0041] Figure 7 Another asymmetric basic road section provided by an embodiment of the present invention is a mixture of expressways and national and provincial roads;

[0042] Figure 8 A schematic diagram of cumulative driving distance provided by an embodiment of the present invention;

[0043] Fig. 9 A basic road section consisting of national and provincial roads provided in an embodiment of the present invention;

[0044] Fig.10 A schematic diagram of a regional trunk line network provided in an embodiment of the present invention;

[0045] Fig.11 A flow chart of another method for estimating the OD quantity of a regional trunk line network provided by an embodiment of the present invention;

[0046] Fig.12 A schematic diagram of functional modules of an OD quantity estimation device for a regional trunk line network provided by an embodiment of the present invention;

[0047] Fig.13 A schematic diagram of the hardware architecture of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] Since highways have ETC data, they can directly obtain the OD distribution of highway ETC. However, there are relatively few traffic detection equipment on local arterial roads and a lack of relevant data, which makes it difficult to obtain the OD distribution of local arterial roads.

[0050] In the existing research on OD estimation, OD back-estimation technology has received important support and rapid development, becoming a very economical and efficient method, providing new means and methods for OD estimation of main roads in various places.

[0051] First, an urban road network traffic flow OD uncertainty estimation method based on vehicle license plate recognition data is proposed. This method is based on vehicle license plate recognition data and integrates the use of computational graphs and Bayesian neural networks to estimate historical and real-time road network traffic flow OD uncertainty. First, a road network model is built according to the generation and distribution rules of traffic flow, and a hierarchical mapping relationship is built based on actual information to reflect the actual distribution of traffic flow. Historical OD data is used as training samples to obtain regional OD uncertainty estimation parameters, which are used as input to estimate real-time traffic flow OD, and indicators are selected to analyze the accuracy and reliability of the estimation results.

[0052] It mainly estimates OD distribution based on historical data. This method obtains the regional historical OD distribution matrix through license plate recognition, GPS data analysis and other methods. The OD matrix obtained by this method is affected by the composition and size of the sample volume, and the estimated OD distribution is greatly affected by some vehicle types and sample characteristics; this method uses actual road network information as input to build the relationship between the levels of the road network model, or uses it as training data to build a road network model to obtain road network parameters, thereby simulating the operation of the road network. However, this method rarely considers the OD transfer and increase caused by the future development of the regional road network, and its OD results may deviate with the development of the city; this method uses real-time section traffic data as input, and historical section traffic and OD distribution as the basis to obtain OD estimation results based on real-time vehicle data. This method has a strong dependence on historical OD distribution.

[0053] Secondly, a provincial highway network OD generation method based on highway toll data and traffic control station data is proposed. This method obtains highway OD data through highway toll stations and highway flow data through traffic distribution; obtains vehicle data on some sections of national and provincial trunk roads through traffic control stations, and uses simulation software to perform OD reverse calculation. By comparing the data of key sections, the model parameters are adjusted to obtain the OD distribution data of national and provincial trunk roads.

[0054] This method can estimate the OD distribution through the OD inverse method according to the traffic volume of the road section. The existing OD inverse method needs to build a traffic model to fit the actual road. The road network model has many influencing factors and cannot accurately reflect the actual road network situation. The road network model is difficult to build and takes a long time. This method uses road section detection data to obtain the OD distribution matrix of national and provincial trunk roads through the OD inverse method. This method is greatly affected by the accuracy of the prior matrix and the inverse method, and there is a problem of unstable OD matrix accuracy. The OD matrix is ​​used to distribute traffic to obtain the section flow, and its distribution results are greatly affected by different traffic distribution methods. This method judges the model fitting by analyzing the road flow and matrix distribution, optimizes the model by adjusting parameters and distribution methods, and obtains regional OD data. The accuracy of the generated OD is judged by comparing the flow of key sections, and its OD distribution results lack robustness and uniqueness.

[0055] The aforementioned existing OD estimation methods are mainly based on the cross-section observed flow data and the OD inverse algorithm to obtain the OD distribution estimation. However, the distribution result of this method is greatly affected by the prior matrix, has high requirements on the integrity of the cross-section road flow, is sensitive to parameters, and the inverse result lacks stability.

[0056] Based on this, an OD quantity estimation method, device and electronic device for a regional trunk road network provided in an embodiment of the present invention can utilize different lane change probability calculation methods corresponding to traffic flow types of various trunk roads in the regional trunk road network to calculate the OD quantity estimation results of the trunk roads of each traffic flow type, thereby ensuring the accuracy of OD quantity estimation of the regional trunk road network.

[0057] To facilitate understanding of this embodiment, a method for estimating the OD amount of a regional trunk line network disclosed in an embodiment of the present invention is first introduced in detail. The method can be applied to controllers such as servers, host computers, and intelligent control devices.

[0058] First, the regional trunk road network is an area between at least one starting point and at least one end point connected and intersected by multiple trunk road paths; the intersection of the paths is a path node; each path node is adjacent to at least one basic road section; the driving flow of the target object on the same basic road section is the cross-sectional flow; the lane change flow between different basic road sections of the target object is the intersection flow; trunk roads include expressways and national and provincial roads (ordinary trunk roads); wherein, each trunk road in the regional trunk road network can be composed of several basic road sections and path nodes between each basic road section; the target object can be understood as a vehicle traveling on the trunk road, and the embodiment of the present invention is explained by taking a vehicle as an example.

[0059] The first intersection detection point or highway toll station of the observation section (the current research section in the basic section) entering the regional trunk road network is taken as the starting point of the vehicle trip, and the last intersection detection point or highway toll station exiting the regional trunk road network is taken as the end point of the vehicle trip. From the vehicle entering to exiting the observation section, the intermediate driving path can be regarded as composed of multiple basic sections, as follows Figure 2 As shown:

[0060] The bold black route is the driving path of multiple vehicles from section i to section i+2. During the driving process, the vehicles passed through multiple path nodes that played the role of lane changing, namely, lane changing nodes i, i+1, and i+2. Therefore, the inventors found that in order to realize the OD estimation of all vehicles entering and exiting the regional trunk road network, firstly, the lane changing probability of all path nodes in the regional trunk road network is calculated, and then the incoming flow at different OD starting points is multiplied by the lane changing probability of each path node passed along the way to the end point, and finally all OD quantities in the regional trunk road network are obtained to realize the estimation of the OD quantity.

[0061] Figure 1 A flow chart of an OD estimation method for a regional trunk line network provided in an embodiment of the present invention.

[0062] Reference Figure 1 , the method comprises the following steps:

[0063] Step S102, determining the flow type of the basic road section adjacent to each path node in the regional trunk road network according to the preset direction from the starting point to the end point;

[0064] The starting point and the end point here are the starting point and the end point in the OD data respectively; a preset direction trend can be formed from the starting point direction to the end point direction in the regional trunk road network; when the target object vehicle drives from the starting point to the end point of the regional trunk road network, each path node in the regional trunk road network flows through the OD traffic flow according to the trend of the preset direction;

[0065] It can be understood that each path node has at least one adjacent basic segment;

[0066] Since the monitoring data sources of expressways and national and provincial roads are different, expressways can obtain ETC data, and estimate the driving path of each vehicle through license plate number, driving location and arrival time, so as to obtain the OD matrix of the group through aggregate statistics. The acquisition of national and provincial road data is generally through the cross-sectional flow traffic detection equipment set on national and provincial roads, which can only obtain the flow data and average speed of all vehicles passing through the location in unit time. Therefore, it is necessary to further distinguish the lane change probability based on the flow types of different roads.

[0067] The traffic types include unknown intersection traffic and known intersection traffic. The detectability of intersection traffic corresponds to several basic road section combination scenarios in practical applications:

[0068] (1) Figure 3 As shown, the basic road segment combination consists of expressways;

[0069] Two cross-modulation stations are set up on expressways J, K and L to detect cross-sectional flow. Figure 3 Each highway j, k, and l is divided into sections i-1 and i by path nodes; the traffic control station set on highway j can detect the cross-sectional flow of section i-1 respectively. and the cross-sectional flow of section i The traffic flow monitoring station set up on highway k can detect the cross-sectional flow of section i-1. and the cross-sectional flow of section i The traffic flow monitoring station set up on highway l can detect the cross-sectional flow of section i-1. and the cross-sectional flow of section i Each path node plays the role of lane change between different highways. Two interchange stations are set between different highways to detect the intersection flow when vehicles drive from one highway to another highway from the path node of the lane change. For example, the intersection flow from highway j to highway k is The flow rate at the intersection when switching from highway k to highway l is The flow rate at the intersection when switching from highway l to highway k is The flow rate at the intersection from highway k to highway j is Therefore, for this type of road segment combination, the traffic flow of all road segments can be obtained, that is, it belongs to the case where the intersection flow is known;

[0070] For highways, its ETC and other collection equipment can detect data including: license plate number, vehicle model, entrance location (entrance toll station code), entrance time, exit location (exit toll station code), exit time, etc.

[0071] (2) A symmetrical basic road section consisting of a mixture of expressways and national and provincial roads. Figure 4 and Figure 5 As shown:

[0072] Figure 4 It can be seen from the basic road segment combinations shown that the expressway k is set between the national and provincial highway j and the national and provincial highway l; Figure 5 It can be seen from the basic road segment combinations shown that the national and provincial highway k is set between the expressway j and the expressway l; Figure 4 and Figure 5 The short line in bold black is used by the traffic station to detect the cross-sectional flow of the corresponding highway and the intersection flow of the highway lane change. The specific flow is similar to the flow diagram of the basic road section combination shown in (1), and will not be repeated here. It can also be seen that the traffic flow of all road sections can be obtained for this road section combination, that is, the intersection flow is also known.

[0073] (3) Asymmetric basic road section combinations consisting of a mixture of expressways and national and provincial roads. Figure 6 and Figure 7 As shown:

[0074] Figure 6 , 7 It can be seen from the basic road section combination shown that the expressway l is located on both sides, while the national and provincial highways j and k are set adjacent to each other; for national and provincial highways, generally no interchange stations are set between national and provincial highways; so Figure 6 , 7 The short-line interchange station indicated by the bold black line is used to detect the cross-sectional flow of each highway, as well as the intersection flow of the expressway l and the national and provincial highway k. The specific flow is similar to the flow diagram of the basic road section combination shown in (1), and will not be repeated here. It can also be seen that the intersection flow between the national and provincial highway j and the national and provincial highway k in this road section combination cannot be obtained, that is, the intersection flow is unknown.

[0075] (4) The basic road section combination consisting of national and provincial roads is composed of Fig. 9 As shown:

[0076] Fig. 9 The intersection stations indicated by the bold black short lines are set up on the left and right sides of each national and provincial highway to detect the cross-sectional flow of each highway. The specific flow is similar to the flow diagram of the basic road section combination shown in (1) and will not be repeated here. It can also be seen that the intersection flow between the national and provincial highways in this road section combination cannot be obtained, that is, the intersection flow is unknown.

[0077] Step S104, estimating the preliminary lane change probability of the path node according to the calculation method corresponding to the flow type of the basic road section adjacent to each path node in the regional trunk road network;

[0078] Here, according to the traffic type of the basic road section adjacent to each path node in the aforementioned embodiment, the corresponding basic road section combination scenario can be subdivided; in order to ensure the accuracy of the calculation of the preliminary lane change probability of the present application, it is necessary to select the corresponding calculation function or model according to the traffic type to realize the estimation of the preliminary lane change probability of the path node;

[0079] Exemplarily, if the intersection flow of the basic road segment adjacent to the path node is unknown, the preliminary lane-changing probability of the path node is calculated according to the lane-changing influencing factors of the path node and the logistic regression model;

[0080] Among them, according to the aforementioned embodiments, it can be known that the asymmetric basic road section combination of a mixture of expressways and national and provincial roads, as well as the scene of the basic road section combination of national and provincial roads, all need to calculate the preliminary lane change probability of each path node in such basic road section combination to other basic road sections according to the logistic regression model; it can be understood that the lane change influencing factors can be obtained from a road data set obtained through methods such as pre-field collection and third-party acquisition; the road data set includes road conditions, traffic conditions and environmental conditions; road conditions include the number of lanes, lane width, lane conditions, etc.; traffic conditions include design speed, headway, road load, travel time, road vehicle type, etc.; environmental conditions include signs and road greening, etc.;

[0081] It should be noted that the preliminary lane change probability is used to train the probability graph model.

[0082] Step S106, input the intersection flow and the speed of the target object into the trained probability graph model to calculate the lane change probability of the path node;

[0083] It should be noted that the traffic flow collected by the traffic control stations on expressways and national and provincial roads is generally collected at a preset frequency, such as 15 minutes. The traffic flow collected by such traffic control stations includes: time, vehicle type, direction, traffic volume, average speed and other data.

[0084] For example, in the basic road section combination scenario in which the two sides of each highway are not equipped with interchange stations in the aforementioned embodiment, there is a situation where the cross-sectional flow cannot be fully known; if in the various basic road section combination scenarios, the interchange stations shown by the bold black short lines are not set on both sides of each highway, then this basic road section combination scenario belongs to the situation where the cross-sectional flow is unknown. In order to further ensure the accuracy of the estimated lane change probability, the probability graph model is trained based on the preliminary lane change probability of each path node obtained in the steps of the aforementioned embodiment at this time, and then the intersection flow and vehicle speed that can be collected in each basic road section combination scenario are input into the trained probability graph model to realize the lane change probability of each path node in this scenario to other basic road sections;

[0085] As an optional embodiment, the probabilistic graphical model may include a conditional random field model CRF.

[0086] Step S108, calculating the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network.

[0087] Here, according to the aforementioned embodiment, different lane change probability calculation methods corresponding to the traffic flow types of various trunk roads in the regional trunk road network are used to calculate the lane change probability of each path node corresponding to the trunk road of each traffic flow type, and then calculate the OD volume estimation result of the regional trunk road network.

[0088] In a preferred embodiment of the actual application, firstly, the preset direction corresponding to the starting point to the end point is determined according to the driving direction of the target object in the regional trunk road network, and then the traffic type of the basic road section adjacent to each path node in the regional trunk road network is obtained according to the road combination scene of the basic road section adjacent to each path node; secondly, the initial lane change probability calculation method of the corresponding path node can be selected according to the traffic type of the basic road section adjacent to each path node. For example, if the intersection traffic of the basic road section adjacent to the path node is unknown, the initial lane change probability of the path node is calculated according to the logistic regression model; if the intersection traffic of the basic road section adjacent to the path node is known, the initial lane change probability is directly calculated, and a probability graph model is trained based on the calculated initial lane change probability of each path node, and the intersection traffic that can be collected and the driving speed of the target object are input into the trained probability graph model to calculate the lane change probability of the path node; by calculating the lane change probability of each path node in the regional trunk road network in this way, the OD amount of each path in the regional trunk road network can be more accurately estimated.

[0089] It is understandable that the basic data of highways is generally ETC data, including basic vehicle attributes (license plate number, vehicle type, etc.), time information (departure time, arrival time), spatial information (departure point, arrival point), etc. By counting the data of vehicles passing through ETC within a certain period of time, the OD table of vehicles with ETC can be obtained, including the starting and ending positions and the number of different vehicle models.

[0090] In actual data acquisition, ETC data does not achieve 100% complete coverage and is not full data, while the OD estimation of highways is an estimation of full data. Therefore, in order to further improve the accuracy of OD estimation of highway-type sections in regional trunk road networks, the ETC data is expanded by combining part of the ETC data and intermodulation data to obtain full highway OD data. The specific steps are as follows:

[0091] Data expansion: Based on the available sample information obtained from data preprocessing, the high-speed OD distribution table is obtained by combining factors such as the proportion of ETC vehicles, day and night rate, and solar-lunar unevenness coefficient.

[0092]

[0093] Among them, H ab is the OD of the expanded sample from toll station a to b; h abis the sampling OD volume from toll station a to b; Da is the proportion of vehicles using ETC for travel at toll station a; e is the day-night rate, f = 24-hour observed traffic volume / daytime observed traffic volume; f and g are the daily and monthly traffic unevenness coefficients respectively; e, f, and g can be obtained through historical traffic volume;

[0094] Traffic allocation: The expanded OD results are used for traffic allocation. The allocation method adopts the random user equilibrium method, and the road resistance function is the BPR function:

[0095]

[0096] Among them, t i is the actual travel time of road section i; t i0 is the free flow travel time of section i; Q is the actual traffic volume of the section; C is the designed traffic capacity of the section; α and β are model parameters.

[0097] Data verification: Compare the distribution results with the traffic detected at the intersection point, correct the OD distribution table, and redistribute the corrected OD distribution results to the road to obtain the allocated traffic on the highway section.

[0098]

[0099] Among them, O i is the actual flow of observed section i; A i Allocate traffic to the observed section i. When the value of the section traffic allocation index GEH is less than the preset value (such as 5), the error is considered to be small. The number of sections whose section traffic allocation index GEH is less than 5 obtained through model allocation is generally considered to be more accurate when the number of sections that meet GEH accounts for more than 60% of the total sections. At this time, the basic section traffic of the highway is calculated based on the current allocated traffic situation.

[0100] d) On this basis, the road section flow can be obtained through model parameter calibration, and the probability of lane change selection at the highway path node can be determined.

[0101] Based on the above embodiments, for flow types where the intersection flow and the cross-sectional flow are known, the following correspondences can be used: Figure 3 , Figure 4 and Figure 5 In the symmetrical basic road segment scenario shown above, which is composed of a mixture of expressways and national and provincial roads, when the basic road segment consists of the above three forms, since the road segment flow of adjacent roads can be obtained from the toll station / interchange station entering and exiting the expressway, the flow of each section and the intersection flow can be obtained; the following method can be used to calculate the preliminary lane change probability of each basic road segment adjacent to each path node change in this basic road segment scenario:

[0102] Exemplarily, if the intersection flow of the basic road segment adjacent to the path node is known, the preliminary lane change probability of the path node is calculated according to the proportional relationship between the intersection flow and the cross-sectional flow of each basic road segment adjacent to the path node.

[0103] For any path node, when the path node is located on a highway, if the vehicle travels from left to right, the left side of each path node is the upstream section of the node, and the right side of each path node is the downstream section of the node; Figure 3 As shown in the figure, the cross-sectional flow and intersection flow of the upstream and downstream sections of each path node of the basic road section composed of this highway are known. When passing through the path node i on the highway k, there are three lane change options at the path node; the vehicle can choose to continue driving on this road, and the corresponding probability is Or enter other highways through the ramp, the corresponding probabilities are Based on the above research, the traffic volume of each road section is known, and the calculation formula for the preliminary lane change probability of the path node is as follows:

[0104]

[0105]

[0106]

[0107] Based on the above steps, the total OD estimate H of the highway is obtained. ab 、Road distribution flow A i and the initial lane change probability of the node on the highway section It should be noted that the parameters in formulas (4)-(6) are all recorded in the aforementioned embodiments and will not be repeated here.

[0108] In some embodiments, Figure 6 , Figure 7 The asymmetric basic road segment scenario shown, which consists of a mixture of expressways and national and provincial roads, cannot detect the flow of each intersection; when the path node i is located on the national and provincial road k, adjacent to the expressway l, there are three lane change options;

[0109] Since there are toll booths / interchange stations between expressway l and national and provincial trunk roads, the intersection flow can be obtained Therefore, the initial lane-changing probability between the national and provincial highway k and the expressway l can be directly calculated according to the formulas (4)-(6) in the above embodiment: and

[0110] Ideally, if detectors are deployed between national and provincial roads, the intersection flow can be obtained. The initial lane-changing probability between national and provincial highways k and j can be directly calculated. The calculation method is the same as that in the previous article, see equations (4) to (6). However, in fact, especially for the upper and lower connecting sections between national and provincial highways, there is generally no intersection point and no detector is deployed, so it is difficult to obtain the above two intersection flows. Therefore, the following method is proposed to obtain the initial lane-changing probability of the path node:

[0111] The basic road traffic types in step S104 may correspond to the following: Figure 6 , Figure 7 The asymmetric basic road segment scenario shown is composed of a mixture of expressways and national and provincial roads, and Fig. 9 The basic road section consists of national and provincial highways; the following method is used to calculate the preliminary lane change probability of each basic road section adjacent to each path node change in this basic road section scenario; where, Figure 6 , Figure 7 The asymmetric basic road section scenario shown is composed of a mixture of expressways and national and provincial roads. Generally speaking, detectors are not deployed between national and provincial roads, and the intersection flow between national and provincial roads cannot be known; in addition, Fig. 9 At least 10 directions of traffic flow data (the directions on both sides of the three roads and the two intersection directions between every two roads) are required to calculate the preliminary lane change probability of each node adjacent to the basic road section. However, under normal circumstances, the intersection flow of the connecting sections between national and provincial roads cannot be obtained, so only 6 cross-sectional flow data can be obtained for calculation; according to the traffic type corresponding to the above two basic road section combination scenarios, step S104 specifically includes:

[0112] Step 2.1), according to the lane-changing influencing factors of the path node, determine the utility of each basic road segment adjacent to the path node;

[0113] Step 2.2), based on the utility of each basic road segment adjacent to the path node change in the logistic regression model, calculate the preliminary lane change probability of each basic road segment adjacent to the path node change.

[0114] Influenced by common driving habits, vehicles will choose roads with faster driving speeds and shorter driving distances when traveling from a starting point to a destination. Therefore, the embodiments of the present invention assume that the lane changing factors considered by the vehicle at the lane changing node are the road type, congestion conditions, and the cumulative driving distance from the destination.

[0115] As shown in the cumulative driving distance Figure 8 As shown, for road section i, the cumulative travel distances from the lane change nodes of the three trunk roads to the final calculated travel destination are It is understandable that the cumulative driving distance can be obtained by summing up the actual research road network.

[0116] Therefore, for any national or provincial highway, the probability of choosing other roads is:

[0117]

[0118]

[0119] Among them, for the set [j, k, l], Θ is the complement of ε, σ, α, β are the parameters to be calibrated, which can be calibrated by actual data; is the utility value of road ε on road section i, that is, the possibility of choosing lane change road ε on road section i, x m are the factors affecting lane change (including road conditions, number of lanes, road speed limit, congestion, road type, and cumulative driving distance, etc.). The Logit model can be used to obtain the probability of road ε choosing other roads on section i, that is, the probability of lane change node on section i.

[0120] In actual applications, it is usually impossible to know the cross-sectional traffic flow data (volume, speed, etc.) of all basic sections in the regional trunk road network, and only some incomplete traffic data are available. The embodiment of the present invention obtains the preliminary lane change node probability for each path node corresponding to the basic section where the interchange station has been deployed through the aforementioned steps, and uses the preliminary lane change node probability corresponding to each path node as the training data of the probability graph model, and introduces the probability graph model to achieve further accurate estimation of the lane change node probability of the remaining basic sections where the interchange station is not deployed, that is, to obtain the lane change node probability of each path node in the regional trunk road network switching to each adjacent basic section. As a preferred embodiment, the conditional random field model in the probability graph model can be used for application; step S106 of the aforementioned embodiment can be implemented by the following steps, including:

[0121] Step 3.1), the intersection flow of the basic road section adjacent to the path node and the speed of the target object are input into the trained probability graph model, and the lane change probability of each basic road section adjacent to each path node in the regional trunk road network is calculated respectively; wherein the trained probability graph model includes the calibrated road type, the number of lanes, and the corresponding weights of the univariate local feature function and the binary transfer feature function.

[0122] Since the lane-changing probability of each path node is related to the traffic flow, speed, and segment attributes of its adjacent basic road segments (upstream and downstream), the initial lane-changing probability of each node can be recorded as As the annotation data of the path node to be estimated for the current lane change probability. At the same time, the input feature vector x is defined as:

[0123]

[0124] X=[x1,…,x i ,…x n ] (10)

[0125]

[0126] in, The flow rate observed at the nearest cross-section observation point upstream of section i speed Road Type and number of lanes The road type and the number of lanes are discrete values, m is the number of upstream and downstream related road sections, and the road type value space is [1, 2, 3, 4], corresponding to expressways, national roads, provincial roads, and ramps, respectively.

[0127] Therefore, input x i The probability value of the output under the condition is:

[0128]

[0129]

[0130] Among them, Z(x) is the normalization factor, which is the sum of all possible values ​​of y; s l (y i ,x,i) is called a univariate local feature function, and the corresponding weight is μ l , It is called the binary transfer feature function, and the corresponding weight is λ k It should be noted that k in formulas (12) and (13) has no physical meaning, and x and y are the independent variable and the dependent variable, respectively. Those skilled in the art can know the specifics from the basic knowledge of the CRF model, which will not be elaborated here.

[0131] By training the conditional random field model, the lane change probability of the path node can be output under the condition of known upstream and downstream section flow, speed, road type and number of lanes.

[0132] For road sections without interchange stations, the flow, speed, road type and number of lanes observed at the upstream and downstream section observation points are input. Since the different types of roads connected to the current road and the upstream and downstream are taken into account, the conditional random field model contains the three different scenarios described above, which can all be uniformly predicted by the model, and the final output is the lane change probability of the node.

[0133] Randomly select training samples and test samples to train the model. During the training process, in order to reflect that some sections of the road have no intermodulation data, they are masked and trained: first, the input sequence is tokenized, and each input element is mapped to the corresponding lane change probability. This usually involves using a vocabulary or character table for tokenization. Set a special marker as a mask marker to indicate the part that needs to be masked, and replace the corresponding input element with the mask marker. Use the input sequence with the mask marker to train the CRF conditional random field model. During the training process, the model will learn how to predict the target variable based on the contextual information, while taking into account the impact of the masked part. Finally, a conditional random field probability model is obtained.

[0134] In some embodiments, the lane change probability of each path node is not completely independent. For example, if a vehicle enters the expressway from the highway via a ramp, the probability of entering the expressway again after returning from the expressway will be reduced; in order to further improve the estimation accuracy of the OD volume of each path in the regional trunk road network, before step S108, for nodes that leave the expressway and then change lanes to the expressway, the lane change probability is corrected by the following steps, specifically including:

[0135] According to the preset direction from the starting point to the end point, if the observed road where the target object has changed direction before the current path node includes a highway, the lane change probability of each basic road segment adjacent to the current path node change is corrected by the preset exponential function exp:

[0136]

[0137]

[0138] The correction based on driving habits is as described in the above equations (14) and (15); wherein h represents the highway section, L represents the vehicle driving path, is the corrected probability of changing lanes to the highway.

[0139] On the basis of the above-mentioned embodiment, the error between the downstream flow calculated by the conditional random field and the actual observed flow should be minimized. In order to further improve the estimation accuracy of the OD amount of each path in the regional trunk road network, an optimization function can be established to correct the lane change probability of the path node; before step S108, the lane change probability of the path node estimated and calculated by the conditional random field is corrected by the following method, including:

[0140] If there is a lane-changing probability of the path node calculated according to the trained probability graph model, then the objective function of the estimated flow of each basic road section adjacent to the path node is determined according to the lane-changing probability of the path node calculated according to the trained probability graph model; the lane-changing probability of the path node calculated according to the trained probability graph model is corrected so that the difference between the estimated flow of each basic road section adjacent to the path node and the intersection flow of each basic road section adjacent to the path node meets the target expectation:

[0141]

[0142]

[0143]

[0144]

[0145] The objective function of the above formula (16) is further solved by genetic algorithm to obtain the corrected lane change probability: in, is the estimated lane change probability before path node correction, is the corrected lane change probability of the path node

[0146] Based on the above-mentioned embodiment, step S108 can calculate the OD value of each path in the regional trunk line network through the following steps, including:

[0147] According to the product of the flow rate corresponding to each starting point in the regional trunk road network and the lane change probability of each basic road section adjacent to each path node change, the OD quantity of the path from each starting point to each end point in the regional trunk road network is calculated;

[0148] The lane-changing probability includes a corrected lane-changing probability and an uncorrected lane-changing probability; that is, both the corrected and uncorrected lane-changing probabilities can be used to estimate the OD quantity of the path from each starting point to each end point in the regional trunk road network through the above steps;

[0149] Specifically, the lane change probability of all path nodes is obtained And multiply the inflow traffic volume of each OD starting point and the probability of all path nodes, and then get the traffic volume of each OD point and the corresponding path. t To D t , the path passed is R t =[r1,r2,…r n ], the corresponding path node is From O t To D t Traffic volume N t :

[0150]

[0151] Then the corresponding OD estimation results of all the regional trunk line networks are:

[0152] ∑ t∈[1,2,…,T] N t (18)

[0153] For the basic section composed of expressways, the embodiment of the present invention pre-processes the expressway ETC data to obtain the expressway sampling OD distribution table, combines the expressway interchange point data and regional characteristics, determines the sample expansion coefficient, obtains the expressway OD distribution table, and on this basis, verifies the data results, and finally obtains the preliminary lane change probability and road flow of each path node of the expressway. For the basic section in the form of a combination of expressways and national and provincial trunk roads, two types of scenarios are proposed according to the location of the expressway. For the basic section of the symmetrical combination, the preliminary lane change probability of the path node can be directly obtained; and for the section lacking interchange points, the scenario of the lane change situation of the typical path node is proposed. With the road data set as input, the Logit model is introduced to calculate the road selection probability, so as to obtain the intersection flow and preliminary lane change probability of the national and provincial trunk roads; on the basis of the aforementioned embodiment, the preliminary lane change probability of all lane change nodes in the trunk road network of the study area is obtained, and it is used as a training sample, and the conditional random field model is further introduced to use the upstream and downstream flow information of the path node at the corresponding moment and the next moment. As input, the estimation of the lane-changing probability of the path node is realized, and the lane-changing probability of the path node based on the random field model is obtained;

[0154] In order to further improve the accuracy of OD estimation, driving habits are used as correction factors to obtain the lane change correction probability of path nodes. Furthermore, the lane change correction probability of path nodes is obtained with the goal of minimizing road flow error. Finally, the input flow data of all O points in the study area are multiplied by the probability of each lane change node to determine the OD distribution of national and provincial trunk roads in the road network.

[0155] The embodiment of the present invention uses highway ETC data, intersection point road detection data and road information data, and uses the Logit model to calculate the preliminary lane change probability of path nodes, introduces conditional random fields to estimate the lane change selection probability of path nodes, and considers driving habits and road flow error factors to correct the probability, so as to scientifically and accurately obtain the highway OD distribution results of national and provincial trunk roads.

[0156] In some embodiments, the present invention also provides another flow chart of a method for estimating the OD amount of a regional trunk line network, such as Fig.11 As shown:

[0157] For highway scenario 1, the ETC data is preprocessed, and then the data is expanded to obtain the OD distribution table, and the road allocation flow and lane change probability are calculated;

[0158] For scenario 2 where national and provincial roads and expressways are mixed, determine whether the expressway is in the middle; if so, calculate the road allocation flow and lane change probability; if not, determine whether detectors are evenly deployed upstream and downstream of the national and provincial roads; if so, calculate the road allocation flow and lane change probability; if not, calculate the lane change probability based on the logistic regression model, and then calculate the road allocation flow and lane change probability;

[0159] For national and provincial highways, determine whether detectors are evenly deployed upstream and downstream of the national and provincial highways; if so, calculate the road allocation flow and lane change probability; if not, calculate the lane change probability based on the logistic regression model, and then calculate the road allocation flow and lane change probability;

[0160] The lane-changing probability of the path nodes in the above three scenarios based on the conditional random field is estimated;

[0161] Consider driving habits to modify the lane change probability of corresponding path nodes;

[0162] The path nodes of downstream traffic (calculated by conditional random fields) are considered to correct the lane-changing probability;

[0163] The OD estimation result is calculated based on the corrected lane changing probability.

[0164] As an optional embodiment, Fig.10 As shown, the bold sections in the road network are expressway sections, the remaining non-bold sections are ordinary national and provincial road sections, the square dotted vertical lines are interchange stations, and the circular dotted vertical lines are the traffic that can be counted based on the ETC data expansion sample.

[0165] First, obtain ETC data, traffic flow data, road network base map data and traffic zone data in the regional trunk road network; among them, ETC data includes expanded sample data, and traffic flow data is data collected by traffic points / stations; traffic zone data can be understood as zone data corresponding to starting points 1, 2 and end points 3, 4 respectively; pre-process the above acquired data (de-duplicate, remove outliers such as null values); determine the estimated OD point according to the traffic zones 1, 2, 3, 4 obtained by division; determine the possible connection path from the starting point to the end point according to the traffic zone and the road network base map; calculate the distance from each node to different end points according to the road network data and length;

[0166] The road types and topological relationships are used to classify the basic road sections, as shown in Table 1:

[0167] Table 1 Basic road section classification table

[0168]

[0169]

[0170] Among them, 1-5, 2-6, 15-4, 17-3, 8-12-15, 14-15, 9-16-17, and 14-17 are all single roads, and there is no lane changing.

[0171] It should be noted that for the lane change nodes in the road network of the embodiment, there are only two alternative sets of lane changes; for example, starting from starting point 1 to node 5, node 5 has two possibilities of lane change node 9 and lane change node 7; if the current node is node 7, node 7 has two possibilities of lane change node 8 and node 10; each path node with lane change behavior has only two alternative sets;

[0172] For the road network in the embodiment, since some sections have no intersection points, their intersection data are unknown (13-16, 16-17, 13-14, 14-15, 6-8). Therefore, the data with known intersection data are used to estimate the probability of lane change nodes on roads with unknown intersection data. Specifically, the conditional random field model is used for calculation as in the steps of the aforementioned embodiment.

[0173] According to the above embodiment, the lane change probability of all nodes is obtained And multiply the probabilities of all lane change nodes between OD points, and then get the traffic volume of OD points and corresponding paths. t To D t , the path passed is R t =[r1,r2,…r n ], the corresponding lane-changing node is From O t To D t Traffic volume:

[0174]

[0175] The corresponding OD estimation table results for all trunk road networks in the study area are shown in Table 2:

[0176] Table 2 OD estimation calculation table

[0177]

[0178] In Table 1 above, due to the different paths to reach the traffic zones, the combination of basic segments is also different, that is, i in each cell in Table 1 above corresponds to a certain road segment of the connecting path between the traffic zones. The OD table is finally estimated by multiplying the lane change node of the basic segment with the input flow of the starting traffic zone.

[0179] In terms of data acquisition, the embodiment of the present invention can also realize OD estimation for highway and ordinary road scenes based on only partial ETC data and intermodulation data, and does not have strong requirements for data integrity. In terms of models, compared with traditional OD inversion technology, there is no need to build a traffic simulation model, and there is no need to rely on the prior matrix of OD inversion. The calculation of lane change probability is realized through the topological relationship between roads and the intermodulation data of adjacent sections, combined with the conditional random field model. At the same time, the embodiment of the present invention takes into account the driving habits and the detection flow of actual intermodulation data to achieve the optimal lane change probability correction, which is more in line with the actual situation. Based on the highway OD estimated by the embodiment of the present invention, the travel path of the vehicle in the highway scene can also be obtained.

[0180] like Fig.12 As shown, an embodiment of the present invention further provides an OD volume estimation device for a regional trunk road network, wherein the regional trunk road network is an area where multiple paths intersect between at least one starting point and at least one end point; the intersection of the paths is a path node; each path node is adjacent to at least one basic road section; the driving flow of the target object on the same basic road section is the cross-sectional flow; the lane change flow between different basic road sections of the target object is the intersection flow; the device comprises:

[0181] A determination module, which determines the flow type of the basic road section adjacent to each path node in the regional trunk road network according to a preset direction from the starting point to the end point; wherein the flow type includes the intersection flow type being unknown and the intersection flow type being known;

[0182] A first estimation module estimates a preliminary lane change probability of each path node according to a calculation method corresponding to a flow type of a basic road section adjacent to each path node in the regional trunk road network; wherein the preliminary lane change probability is used to train a probability graph model;

[0183] A second estimation module inputs the intersection flow and the speed of the target object into a trained probability graph model to calculate the lane change probability of the path node;

[0184] A calculation module calculates the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network.

[0185] In some embodiments, the first estimation module is also used to calculate the preliminary lane changing probability of the path node according to the proportional relationship between the intersection flow and the cross-sectional flow of each basic road segment adjacent to the path node if the intersection flow of the basic road segment adjacent to the path node is known; if the intersection flow of the basic road segment adjacent to the path node is unknown, calculate the preliminary lane changing probability of the path node according to the lane changing influencing factors of the path node and a logistic regression model.

[0186] In some embodiments, the first estimation module is further used to determine the utility of each basic road segment adjacent to the path node change based on the lane change influencing factors of the path node; and calculate the preliminary lane change probability of each basic road segment adjacent to the path node change based on the utility of each basic road segment adjacent to the path node change in a logistic regression model.

[0187] In some embodiments, the second estimation module is also used to input the intersection flow and the speed of the target object of the basic road section adjacent to the path node into the trained probability graph model, and respectively calculate the lane change probability of each basic road section adjacent to each path node in the regional trunk road network; wherein the calibrated road type, number of lanes, and corresponding weights of the univariate local feature function and the binary transfer feature function.

[0188] In some embodiments, before the step of calculating the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network, the device also includes a correction module; the correction module is used to correct the lane change probability of each basic road segment adjacent to the current path node according to the preset direction along the starting point to the end point, if the observed road where the target object has changed direction before the current path node includes a highway, using a preset exponential function.

[0189] In some embodiments, before the step of calculating the OD amount of each path in the regional trunk road network according to the lane changing probability of each path node in the regional trunk road network, the correction module is also used to, if there is a lane changing probability of the path node calculated according to the trained probability graph model, determine the objective function of the estimated flow of each basic road segment adjacent to the path node according to the lane changing probability of the path node calculated according to the trained probability graph model; and correct the lane changing probability of the path node calculated according to the trained probability graph model so that the difference between the estimated flow of each basic road segment adjacent to the path node and the intersection flow of each basic road segment adjacent to the path node meets the target expectation.

[0190] In some embodiments, the calculation module is used to calculate the OD value of the path from each starting point to each end point in the regional trunk road network based on the product of the flow corresponding to each starting point in the regional trunk road network and the lane change probability of each basic road segment adjacent to the change of each path node.

[0191] An embodiment of the present invention provides an electronic device for implementing an electronic device. In this embodiment, the electronic device may be, but is not limited to, a personal computer (PC), a laptop computer, a monitoring device, a server, or other computer device with analysis and processing capabilities.

[0192] As an exemplary embodiment, see Fig.13 The electronic device 110 includes a communication interface 111, a processor 112, a memory 113 and a bus 114. The processor 112, the communication interface 111 and the memory 113 are connected via the bus 114. The memory 113 is used to store a computer program that supports the processor 112 to execute the method. The processor 112 is configured to execute the program stored in the memory 113.

[0193] The machine-readable storage medium mentioned in this article can be any electronic, magnetic, optical or other physical storage device that can contain or store information, such as executable instructions, data, etc. For example, the machine-readable storage medium can be: RAM (Radom Access Memory), volatile memory, non-volatile memory, flash memory, storage drive (such as hard disk drive), any type of storage disk (such as CD, DVD, etc.), or similar storage medium, or a combination thereof.

[0194] The non-volatile medium may be a non-volatile memory, a flash memory, a storage drive (such as a hard drive), any type of storage disk (such as a CD, DVD, etc.), or a similar non-volatile storage medium, or a combination thereof.

[0195] It can be understood that the specific operation methods of each functional module in this embodiment can refer to the detailed description of the corresponding steps in the above method embodiment, and will not be repeated here.

[0196] The computer-readable storage medium provided in the embodiment of the present invention stores a computer program. When the computer program code is executed, the method described in any of the above embodiments can be implemented. For specific implementation, please refer to the method embodiment, which will not be described in detail here.

[0197] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0198] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0199] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0200] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the aforementioned embodiments, those of ordinary skill in the art should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the aforementioned embodiments within the technical scope disclosed by the present invention, or can easily conceive of changes, or make equivalent replacements for some of the technical features therein. Such modifications, changes or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the protection scope of the present invention.

Claims

1. A method for estimating the OD of a regional trunk line network, characterized in that: The regional trunk road network is an area where multiple paths intersect between at least one starting point and at least one end point; the intersection of the paths is a path node; each path node is adjacent to at least one basic road segment; The driving flow of the target object on the same basic road section is the cross-sectional flow; The lane-changing flow between different basic sections of the target object is the intersection flow; The method comprises: Determine the flow type of the basic road section adjacent to each path node in the regional trunk road network according to the preset direction from the starting point to the end point; wherein the flow type includes the intersection flow unknown and the intersection flow known; According to the calculation method corresponding to the traffic type of the basic road section adjacent to each path node in the regional trunk road network, the preliminary lane change probability of the path node is estimated; wherein the preliminary lane change probability is used to train the probability graph model; Inputting the intersection flow and the speed of the target object into a trained probability graph model to calculate the lane change probability of the path node; Calculating the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network; The step of estimating the preliminary lane change probability of each path node according to the calculation method corresponding to the flow type of the basic road section adjacent to each path node in the regional trunk road network comprises: If the intersection flow of the basic road section adjacent to the path node is known, the preliminary lane-changing probability of the path node is calculated according to the proportional relationship between the intersection flow and the cross-sectional flow of each basic road section adjacent to the path node; If the intersection flow of the basic road section adjacent to the path node is unknown, the preliminary lane changing probability of the path node is calculated according to the lane changing influencing factors of the path node and the logistic regression model.

2. The OD estimation method of the regional trunk line network according to claim 1 is characterized in that: If the intersection flow of the basic road section adjacent to the path node is unknown, the step of estimating the preliminary lane change probability of the path node according to the lane change influencing factors of the path node and the logistic regression model includes: Determining the utility of each basic road segment adjacent to the lane change of the path node according to the lane change influencing factors of the path node; Based on the utility of each basic road segment adjacent to the path node change in a logistic regression model, a preliminary lane change probability of each basic road segment adjacent to the path node change is calculated.

3. The OD estimation method of the regional trunk line network according to claim 1, characterized in that: The step of inputting the intersection flow and the speed of the target object into the trained probability graph model to calculate the lane change probability of the path node includes: The intersection flow and the speed of the target object of the basic road section adjacent to the path node are input into the trained probability graph model, and the lane change probability of each basic road section adjacent to each path node in the regional trunk road network is calculated respectively; wherein the trained probability graph model includes the calibrated road type, the number of lanes, and the corresponding weights of the univariate local feature function and the binary transfer feature function.

4. The OD estimation method of a regional trunk line network according to claim 1, characterized in that: Before the step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network, the method further includes: According to the preset direction from the starting point to the end point, if the observed road on which the target object has changed direction before the current path node includes a highway, the lane change probability of each basic road segment adjacent to the current path node is corrected by a preset exponential function.

5. The OD estimation method of a regional trunk line network according to claim 1, characterized in that: Before the step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network, the method further includes: If there is a lane changing probability of the path node calculated according to the trained probability graph model, then the objective function of the estimated flow rate of each basic road segment adjacent to the path node is determined according to the lane changing probability of the path node calculated according to the trained probability graph model; the lane changing probability of the path node calculated according to the trained probability graph model is corrected so that the difference between the estimated flow rate of each basic road segment adjacent to the path node and the intersection flow rate of each basic road segment adjacent to the path node meets the target expectation.

6. The OD estimation method of a regional trunk line network according to claim 1, characterized in that: The lane change probability includes a corrected lane change probability and an uncorrected lane change probability; The step of calculating the OD amount of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network comprises: The OD amount of the path from each starting point to each end point in the regional trunk road network is calculated according to the product of the flow corresponding to each starting point in the regional trunk road network and the lane change probability of each basic road section adjacent to each path node change.

7. An OD estimation device for a regional trunk line network, characterized in that: The regional trunk road network is an area where multiple paths intersect between at least one starting point and at least one end point; the intersection of the paths is a path node; each path node is adjacent to at least one basic road segment; The driving flow of the target object on the same basic road section is the cross-sectional flow; The lane-changing flow between different basic sections of the target object is the intersection flow; The device comprises: A determination module, which determines the flow type of the basic road section adjacent to each path node in the regional trunk road network according to a preset direction from the starting point to the end point; wherein the flow type includes the intersection flow unknown and the intersection flow known; A first estimation module estimates a preliminary lane change probability of each path node according to a calculation method corresponding to a flow type of a basic road section adjacent to each path node in the regional trunk road network; wherein the preliminary lane change probability is used to train a probability graph model; A second estimation module inputs the intersection flow and the speed of the target object into a trained probability graph model to calculate the lane change probability of the path node; A calculation module, which calculates the OD value of each path in the regional trunk road network according to the lane change probability of each path node in the regional trunk road network; The first estimation module is also used to calculate the preliminary lane changing probability of the path node according to the proportional relationship between the intersection flow and the cross-sectional flow of each basic road section adjacent to the path node if the intersection flow of the basic road section adjacent to the path node is known; if the intersection flow of the basic road section adjacent to the path node is unknown, calculate the preliminary lane changing probability of the path node according to the lane changing influencing factors of the path node and a logistic regression model.

8. An electronic device, characterized in that: The method comprises a memory, a processor, and a program stored in the memory and capable of being run on the processor, wherein the processor implements the method according to any one of claims 1 to 6 when executing the program.

9. A computer-readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed, the method according to any one of claims 1 to 6 is implemented.

Images