OD backstepping method based on ETC door frame flow data

Abstract

The invention discloses an OD backstepping method based on ETC portal flow data, solves the problem that an existing OD backstepping method is difficult to reflect traffic mobility, and belongs to the field of traffic engineering and intelligent traffic systems. The method comprises the following steps: determining a research target as region-level OD backstepping, city-level OD backstepping or expressway-level OD backstepping; dividing each composition area related to the research target into a direct traffic influence area and an indirect traffic influence area according to the influence degree; dividing the direct traffic influence area and the indirect traffic influence area into traffic areas according to administrative area units and expressway sections; acquiring flow data of the selected ETC door frames in real time to obtain the hour traffic volume of each ETC door frame, and matching the obtained hour traffic volume to the current road network in combination with the positions of the ETC door frames to obtain the road network section flow; taking the obtained traffic area as a starting point and an ending point, selecting the road network section flow of the ETC portal frame for OD backstepping, and carrying out OD matrix backstepping; and verifying the reversely deduced OD matrix.

Description

Technical Field

[0001] The invention relates to an OD inverse method based on ETC gantry flow data, and belongs to the field of traffic engineering and intelligent transportation systems. Background Art

[0002] With the continuous growth of traffic demand, especially the continuous expansion of highway networks, obtaining regional OD and predicting traffic flow on road networks have become important issues in traffic management and planning. As one of the core data in traffic flow analysis, the OD (Origin-Destination) matrix can reflect the distribution of travel demand in a specific area and is a key tool in the fields of traffic planning, traffic management, and traffic safety warning. However, the existing OD matrix inversion methods face multiple challenges, especially in terms of real-time and regional hierarchical adaptability.

[0003] Traditional OD reverse prediction methods mostly rely on static traffic flow data and cannot effectively respond to rapid changes in traffic flow, especially in the event of emergencies, weather changes, or traffic accidents. They lack the ability to make real-time adjustments and corrections, resulting in poor timeliness of prediction results. In addition, OD reverse prediction in existing technologies is usually limited to a single level and lacks the ability to consider differences in traffic flow between different regions. Traditional methods often perform reverse prediction based on the city level or a single road section, which makes it difficult to fully reflect traffic mobility at the regional or highway level. Summary of the invention

[0004] In view of the problem that the existing OD reverse estimation method is difficult to reflect the traffic fluidity, the present invention provides an OD reverse estimation method based on ETC gantry flow data.

[0005] The present invention provides an OD reverse estimation method based on ETC gantry flow data, comprising:

[0006] S1. Determine the research target as regional-level OD inversion, city-level OD inversion or highway-level OD inversion;

[0007] Regional-level OD back-estimation is used to analyze traffic flows between multiple cities or regions;

[0008] City-level OD back-calculation is used to analyze traffic flows within the city;

[0009] Highway-level OD inversion is used to analyze traffic flow between interchanges of two-way lanes on highways;

[0010] S2. Divide the component areas involved in the research target into direct traffic impact areas and indirect traffic impact areas according to the degree of impact;

[0011] S3. Divide the direct traffic impact area and the indirect traffic impact area into traffic zones according to administrative units and highway sections;

[0012] S4. Select ETC gantry;

[0013] S5. Obtain the flow data of the selected ETC gantry in real time, and after processing, obtain the hourly traffic volume of each ETC gantry. According to the position of the ETC gantry, match the obtained hourly traffic volume to the existing road network to obtain the road network cross-sectional flow;

[0014] S6, taking the traffic area in S3 as the starting and ending points, selecting the road network section flow of the ETC gantry for OD reverse calculation, and performing OD matrix reverse calculation;

[0015] S7. Assign the inversely derived OD matrix to the road network to obtain the road network cross-sectional flow, and extract the road network cross-sectional flow of the section corresponding to the ETC gantry used for verification from the road network cross-sectional flow, and compare it with the actual road network cross-sectional flow of the section corresponding to the ETC gantry used for verification. If the accuracy requirement is met, end; otherwise, go to S6 and reselect the road network cross-sectional flow of the ETC gantry used for OD inversion.

[0016] Preferably, S2 comprises:

[0017] For regional OD back-calculation, the cities and counties involved in the research target are divided into direct impact areas, and the areas outside the direct impact areas where the research target involves vehicles traveling are considered indirect impact areas;

[0018] For city-level OD backcasting, all cities involved in the research target are regarded as direct impact areas, and cities adjacent to the direct impact areas are classified as indirect impact areas;

[0019] For the highway-level OD back-calculation, each upstream and downstream section of the highway involved in the research target is divided into the direct impact area, and the smallest administrative unit connected to the highway entrances and exits and each interchange is regarded as the indirect impact area.

[0020] Preferably, S3 includes:

[0021] For the regional OD back-calculation, for the directly affected area, the traffic zone is divided at the district or county level, or at the township, town, or village level; for the indirectly affected area, the traffic zone is divided using an administrative unit one level larger than the directly affected area;

[0022] For the city-level OD backcasting, for the directly affected areas, the traffic zones are divided according to the municipal districts, or the street-level traffic zones are divided with important traffic nodes and roads as the centroids; for the indirectly affected areas, the traffic zones are divided according to the municipal level;

[0023] For highway-level OD back-calculation, for the direct impact area, each up and down section on the highway involved in the research target is divided into a traffic zone; for the indirect impact area, the districts, counties or villages directly connected to the highway entrances and exits and interchanges are divided into traffic zones.

[0024] Preferably, S5 includes:

[0025] Obtain the 15-minute traffic data of the selected ETC gantry in real time, convert the traffic of different models into PCU traffic through the vehicle conversion coefficient, sum up to get the PCU flow rate of each ETC gantry, and process it into the hourly traffic volume of each ETC gantry;

[0026] Match the location of the ETC gantry with the road network, assign the corresponding hourly flow data to the corresponding ETC gantry, and obtain the current road network cross-sectional flow.

[0027] For S5, another method is provided, including:

[0028] S51, obtaining the selected ETC gantry traffic data, the data time range is one year;

[0029] S52. For each ETC gantry, convert the flow data of each vehicle type into flow in pcu according to the respective vehicle conversion coefficients, and sum the pcu flows of each vehicle type to obtain the total flow in pcu of each gantry;

[0030] S53, for each gantry, the total traffic volume is divided by 365 days to obtain the annual average daily traffic volume of each ETC gantry;

[0031] S55. For each ETC gantry, the hourly traffic volume of each gantry is obtained based on the annual average daily traffic volume and the peak hour ratio;

[0032] S56. Match the location of the ETC gantry with the road network, assign the corresponding hourly flow data to the corresponding ETC gantry, and obtain the current road network cross-sectional flow.

[0033] Preferably, S6 includes:

[0034] S61, set the initial values ​​of k, r, and w, where the initial value of k is 1, and the observed values ​​of the flow of each section are obtained according to the flow of the road network section.

[0035]

[0036] S62, using the capacity-limited iterative balance method to calculate the estimated flow value Q of each road section a :

[0037] Q a=[Q a1 , Q a2 ,···,Q al ] T

[0038] in, The probability of selecting section a for the distribution from area i to area j, p i represents the generation index of zone i, A j is the attraction index of zone j, t ij is the impedance from area i to area j, where impedance is time or distance;

[0039] S63, according to Q a and Update the estimate of k:

[0040] S64, judging whether k converges, if so, proceeding to S64, if not, proceeding to S62;

[0041] S65. Calculate variance σ 2 :

[0042]

[0043] Judgment 2 Has it reached the minimum? If so, end. If not, change the values ​​of r and w and go to S1.

[0044] According to S61 to S65, the determined k, r get q ij :

[0045]

[0046] q ij It is the element of the OD matrix, indicating the OD distribution from area i to area j.

[0047] Preferably, S4 includes:

[0048] For regional OD back-propagation, all ETC gantries in the direct impact area are considered as candidates, and some ETC gantries in the indirect impact area are included as candidates;

[0049] For city-level OD backpropagation, all ETC gantries in the direct impact area will be considered as alternatives;

[0050] For highway-level OD backpropagation, all ETC gantries in the traffic area are considered as candidates.

[0051] Preferably, in S6, the road network cross-sectional flow of the ETC gantry selected for OD reverse thrust includes:

[0052] According to the following principles, the network cross-sectional flow of the ETC gantries among the candidates is selected for OD reverse estimation, and the network cross-sectional flow of the remaining ETC gantries is used for verification;

[0053] Principle 1: Determine the number of OD point pairs covered by each ETC gantry, and then select several gantries that cover a large number of OD point pairs;

[0054] Principle 2: Avoid selecting ETC gantries with highly correlated traffic data;

[0055] Principle 3: For a specific OD point pair, select the gantry that can capture the maximum flow.

[0056] Principle 4: Among multiple alternative gantries, those gantry combinations that can capture the maximum flow should be selected.

[0057] As a preferred method, the method of allocating the inversely derived OD matrix to the road network is:

[0058] Establish the objective function:

[0059]

[0060] The constraints are:

[0061]

[0062] Where: x a is the traffic flow on road section a, t a is the traffic impedance on section a, t a (x a ) is the flow rate x on section a a is the traffic impedance function of the independent variable, f k rs is the flow of the kth path between ODs with the departure point r and the destination point s, If the segment a belongs to the kth path between ODs with the departure point r and the destination point s, then δ=1, otherwise δ=0;

[0063] Under the constraint conditions, the objective function is solved to obtain the inverse OD matrix distributed to the road network to obtain the cross-sectional flow of the road network.

[0064] The present invention effectively optimizes the problem of reverse prediction accuracy and the difficulty of real-time reverse prediction in the prior art by introducing a real-time dynamic correction mechanism, regional hierarchical OD reverse prediction and an accurate traffic zone division method, and has significant advantages in traffic flow prediction accuracy, adaptability and management efficiency. The following are the main advantages and effects of the present invention:

[0065] The present invention can dynamically correct the OD matrix by acquiring ETC gantry flow data in real time and combining it with the traffic flow monitoring system to reflect changes in traffic conditions in real time. Compared with the traditional static OD matrix method, this mechanism can effectively improve the prediction accuracy, especially during peak traffic hours and emergencies. Real-time correction can significantly reduce prediction errors and ensure the timeliness and accuracy of traffic management decisions.

[0066] The present invention adopts a multi-level OD back-calculation model, which can be flexibly adjusted according to different traffic management needs. OD back-calculation at different levels (city level, regional level and highway level) is suitable for traffic management tasks of different scales, from traffic analysis in a large area to traffic prediction on a specific highway, and can accurately meet the needs. This hierarchical design improves the applicability of the OD matrix and avoids the limitations of a single-level model.

[0067] The traffic zone division method proposed in this invention scientifically and reasonably divides the traffic zones within the affected area, and especially accurately predicts the impact of newly built expressways on surrounding areas. By refining the division of traffic zones, the regional division accuracy during OD back-calculation is improved, ensuring a high degree of matching between traffic flow data and regional needs. This method greatly reduces the accuracy deviation caused by traditional rough division, especially in complex traffic networks and newly built expressway traffic flow prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0070] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0071] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0072] The OD inversion method based on ETC gantry flow data of this embodiment includes:

[0073] Step 1: Determine the research target as regional-level OD inversion, city-level OD inversion or highway-level OD inversion;

[0074] According to the research purpose, the design of OD back-propagation can be divided into different levels such as regional level, city level and highway level. The method of selecting the back-propagation level is that for regional-level OD back-propagation, it covers the traffic flow between multiple cities or regions, which is usually suitable for cross-city traffic analysis; for city-level OD back-propagation, the focus is on analyzing the traffic flow within the city, which is suitable for relatively local traffic prediction; for highway-level OD back-propagation, it focuses on the traffic flow distribution between the interchanges of the two-way lanes of the highway, which is particularly suitable for refined traffic prediction of the highway network.

[0075] When selecting the level, it is also necessary to combine the research objectives and the spatial distribution characteristics of traffic flow. The research target of this implementation method is the newly built Harbin-Zhaoqing Expressway. Since the Harbin-Zhaoqing Expressway connects Harbin and Daqing, the traffic flow through the region is complex and changeable. Therefore, the Harbin-Zhaoqing Expressway is not only affected by several surrounding cities, but also by the transportation network and regional socio-economic activities within the entire Heilongjiang Province. Therefore, for the Harbin-Zhaoqing Expressway, the research object of this implementation method, the OD inversion level selected is the regional level, so as to comprehensively consider the impact of the traffic flow of the entire region on the traffic flow of the Harbin-Zhaoqing Expressway.

[0076] Step 2: Divide the component areas involved in the research target into direct traffic impact areas and indirect traffic impact areas according to the degree of impact;

[0077] After completing the regional hierarchical classification and clarifying the purpose and scope of OD back-calculation, the traffic impact area is divided.

[0078] The traffic impact area of ​​the research target refers to the area where the traffic conditions are significantly improved after the research target is built, the time and space distance between regions is relatively shortened, and passenger and freight transportation is more convenient and economical, thereby accelerating the social and economic development of the region. According to the degree of impact, the impact area can be divided into direct impact area and indirect impact area.

[0079] The direct impact area usually has the following characteristics:

[0080] (1) The implementation of the research objectives will bring significant benefits to the socio-economic development of the region.

[0081] (2) Most of the traffic volume undertaken by the research targets comes from these areas, that is, the traffic sources or concentration points are mostly located in these areas.

[0082] (3) After the research objectives are implemented, roads or other modes of transportation in these areas will be significantly diverted, and traffic conditions will be greatly improved.

[0083] (4) From a geographical point of view, the area is generally very close to the research target or the research target directly passes through the area.

[0084] For the OD reverse calculation at the regional level, the cities, counties and other administrative units that the research target directly passes through are divided into direct impact areas; and the range of vehicles traveling in the research target outside the direct impact area is divided into indirect impact areas. For the OD reverse calculation at the city level, all cities involved in the research target are divided into direct impact areas, and cities directly adjacent to the direct impact area are divided into external indirect impact areas. For the OD reverse calculation at the highway level, the upstream and downstream sections of the highway involved in the research target are divided into direct impact areas, and the smallest administrative unit connected to the highway entrances and exits and each interchange is divided into indirect impact areas.

[0085] The purpose of OD backcasting in this implementation method is to predict the traffic flow of the Harbin-Zhaoqing Expressway in the future planning characteristic year. It is necessary to include as many areas as possible that have an impact on the traffic flow of the Harbin-Zhaoqing Expressway, so the OD backcasting level of the Harbin-Zhaoqing Expressway is set as the regional level. According to the location of the Harbin-Zhaoqing Expressway and based on the above-mentioned research target impact zone division principle, Harbin, Daqing, and Suihua are divided into direct impact zones, and other cities and Jilin Province are divided into indirect impact zones.

[0086] Step 3: Divide the direct traffic impact area and the indirect traffic impact area into traffic zones according to administrative units and highway sections;

[0087] Based on the purpose of OD back-calculation and the characteristics of the research objectives, after completing the division of traffic impact areas, the traffic impact areas are further divided into traffic areas so that OD back-calculation can be performed based on traffic areas.

[0088] For regional OD backcasting, because the overall area of ​​OD backcasting is usually large, or further traffic volume forecasting is required, administrative district units that are convenient for collecting economic and population data are used as traffic zones. For direct impact areas, traffic zones are usually divided at the district and county levels. If a more detailed OD distribution is required, the traffic zones of the direct impact areas can also be divided at the township, town or even village levels. For indirect impact areas, administrative district units that are one level larger than the direct impact areas are used for traffic zone division. For example, if the direct impact area is divided into traffic zones at the district and county levels, the indirect impact area is divided into traffic zones at the city level.

[0089] For city-level OD back-calculation, the purpose of OD back-calculation of the research target is generally to analyze the traffic flow within the city or between cities, so the division of traffic zones should be based on the administrative divisions of the city and the nodes where traffic flow occurs. For the direct impact area, the traffic zone can be divided according to the city district. If the research target requires a more accurate OD distribution, the important traffic nodes and roads can be used as the centroid to divide the street-level traffic zone. For the indirect impact area, the city adjacent to the direct impact area and whose traffic flow is affected is the indirect impact area, which is regarded as an external community and the traffic zone is divided according to the city level.

[0090] For highway-level OD inversion, the purpose of OD inversion of the research target is generally to obtain the up-and-down traffic volume of each section of the highway, and the in-and-out traffic volume of highway entrances and exits and interchanges. For the direct impact area, each up-and-down section on the highway involved in the research target is divided into a traffic zone. For the indirect impact area, the counties or villages directly connected to the highway entrances and exits and interchanges are divided into traffic zones.

[0091] For this implementation, as a regional-level OD back-calculation, the traffic zones within the direct impact area are divided into county and district levels, and the indirect impact area is divided into prefecture-level city level and adjacent province level. According to the above principles, a total of 47 traffic zones are divided, as shown in Table 1.

[0092] Table 1 Traffic zone division list

[0093]

[0094]

[0095]

[0096] Step 4. Select ETC gantry:

[0097] For different regional hierarchical classifications, ETC gantries of different ranges and locations should be selected as the source of highway section flow data. The method for selecting ETC gantries is given as follows. The method for selecting the range of ETC gantries is that for regional-level OD reverse calculation, the ETC gantries on all highways in the direct impact area are used as alternatives, and the ETC gantries in the indirect impact area are included through certain selection; for city-level OD reverse calculation, all ETC gantries in the involved cities are used as alternatives; for highway-level OD reverse calculation, all ETC gantries in the traffic area are used as alternatives.

[0098] Step 5: Obtain the traffic data of the selected ETC gantry in real time, and after processing, obtain the hourly traffic volume of each ETC gantry. According to the position of the ETC gantry, match the obtained hourly traffic volume to the existing road network to obtain the road network section flow:

[0099] After completing the traffic zone division, the work of the OD back-estimation preparation phase at the GIS surface layer is basically completed. To further carry out the work at the GIS line layer, the first step is to collect ETC gantry traffic data, clean and process it into the data type required by the research target.

[0100] Obtain the flow data of the selected ETC gantry in real time, convert the flow of different models into pcu flow through the vehicle conversion coefficient, sum up to obtain the pcu flow rate of each ETC gantry, and process it into the hourly traffic volume of each ETC gantry; match the location of the ETC gantry with the road network, assign the corresponding hourly flow data to the corresponding ETC gantry, and obtain the current road network cross-sectional flow. When using this method for real-time OD reverse projection, the real-time ETC gantry flow data obtained is generally short-term road section flow data every 15 minutes, which is converted into hourly traffic volume for OD reverse projection, and the formula for converting it into hourly flow rate is:

[0101] Q 60 =4×Q 15 ×PHF 15

[0102] In the formula, Q 60 ——Hourly flow rate

[0103] Q 15 ——ETC gantry measured flow rate within 15 minutes

[0104] PHF 15 ——Hourly coefficient: use the average hourly coefficient of the area corresponding to each hour. If there is no historical data of hourly coefficient, a road section with strong representativeness should be selected for actual measurement.

[0105] According to Ha Zhao's research objectives, the processing process of this implementation method is as follows:

[0106] Collect data. Data is collected based on the ETC gantry range inferred from the OD of different levels in step one, and all ETC gantry traffic data that can be obtained within the required range are collected. The data time range is one year. For the research objectives of Hazhao, daily traffic volume data of up and down toll-collecting vehicles at 212 gantries on the expressway network in Heilongjiang Province from January 1, 2023 to December 31, 2023 were collected. The toll-collecting vehicle types at the gantries include 4 types of passenger cars (Class 1 to Class 4 passenger cars), 6 types of trucks (Class 1 to Class 6 trucks), and 6 types of special operation vehicles (Class 1 to Class 6 special operation vehicles).

[0107] Clean the data. The data type of the traffic data of each ETC gantry in the highway network is generally the statistical data of the traffic of each vehicle type on a daily or monthly basis. Abnormally small values ​​often appear in these data and need to be removed. They cannot be used as the data source for calculating the traffic of the road network section.

[0108] Process the traffic data into the required data type.

[0109] The data processing steps are as follows:

[0110] Step 51, obtaining the flow data of each gantry from January to December in a year;

[0111] Step 52: For each gantry, the flow data of each vehicle type obtained in step 51 is converted into flow in PCU according to the respective vehicle conversion coefficients, and the PCU flows of each vehicle type are summed to obtain the total flow in PCU of each gantry;

[0112] Step 53: For each gantry, divide the total flow obtained in step 2 by 365 days to obtain the AADT of each ETC gantry, i.e., the annual average daily traffic volume (pcu / d). If the flow data of some months are missing when calculating the total flow, the average value is calculated based on the actual number of days.

[0113] Step 54: For each gantry, calculate the peak hour traffic volume. Take the peak hour ratio as 10% (the peak hour ratio refers to the percentage of peak hour traffic volume to the daily traffic volume of the day), and multiply the AADT obtained in step 3 by the peak hour ratio to obtain the hourly traffic volume of each gantry.

[0114] The vehicle conversion coefficients of each vehicle type in the above step 52. The vehicle type in the traffic data of the ETC gantry is classified according to the toll vehicle type. According to the transportation industry standards and charging standards, the classification of toll vehicle types is shown in Table 2. By comparing the fare vehicle types in Table 2 and the classification of representative car models (see Table 3), the conversion coefficients when the toll vehicle type is converted to a passenger car are determined, see Table 4.

[0115] Table 2 Classification of vehicle types for toll road vehicle tolls

[0116]

[0117]

[0118] Table 3 Highway route design specifications, classification of representative vehicle models

[0119] Representative car models illustrate Passenger car Passenger cars with ≤19 seats and trucks with a load capacity of ≤2t Mid-size car Passenger cars with more than 19 seats and trucks with a load capacity of 2t and less than 7t Large car Trucks with a load capacity of 7t or less and a load capacity of ≤20t Car train Trucks with a load capacity of >20t

[0120] Table 4 Conversion coefficients when charging vehicle types are converted to passenger cars

[0121]

[0122]

[0123] Step 55: Obtain the current motor vehicle cross-sectional flow of the expressway network

[0124] After obtaining the hourly traffic volume data of the ETC gantry in step 53, it is necessary to further match these flow data to the existing road network to obtain the road network section flow.

[0125] Match the location of the ETC gantries with the road network, that is, determine the location of these ETC gantries in the road network. In the process of researching the target, it was found that when collecting ETC gantry information, there is usually only a single ETC gantry number or name list and its corresponding traffic data, or the absolute position of the ETC gantry such as longitude and latitude, but there is no important relative position of the ETC gantry in the road network. To perform OD reverse inference of this method, the cross-sectional traffic of the road network is required. Without matching the ETC gantry position with the road network, the cross-sectional traffic of the road network cannot be obtained, and OD reverse inference cannot be performed. For the research goal of Hazhao, by searching the names of toll stations on the road network, it was found that the naming of ETC gantries basically adopted the names of toll stations at both ends of the high-speed section. Therefore, using software such as AutoNavi that contains road information, all ETC gantries are determined in the road network through manual retrieval.

[0126] After completing the location matching of the ETC gantry and the road network, the corresponding peak hour traffic data is also assigned to the corresponding ETC gantry. The current motor vehicle cross-sectional traffic of the expressway network is shown in Table 5.

[0127] Table 5 Traffic volume of ETC gantry sections in the current expressway network (pcu / d)

[0128]

[0129]

[0130] Table 5

[0131]

[0132]

[0133] Table 5

[0134]

[0135]

[0136]

[0137] Step 6. Take the traffic area in step 3 as the starting and ending points, select the road network cross-sectional flow of the ETC gantry used for OD reverse estimation, and perform OD matrix reverse estimation. After obtaining the motor vehicle cross-sectional flow of the existing expressway network, in order to verify the accuracy of the OD reverse estimation method, some gantries need to be used for OD reverse estimation, and the remaining gantries are used for reverse estimation accuracy verification.

[0138] In order to ensure the accuracy of OD reverse thrust and the persuasiveness of verification at the same time, it is necessary to balance the number of gantries for reverse thrust and the number of gantries used for accuracy verification, and design experiments to obtain the above balance ratio. The test content is to use different proportions of ETC gantries for OD reverse thrust, and design 50%, 75%, 85%, and 90% of the gantry flow data for OD reverse thrust, while the remaining gantry flow data is used for verification. It is found that when 85% of the ETC gantry flow is used for reverse thrust, the accuracy of the reverse thrust result is better, and the proportion of ETC gantries used for verification is appropriate and persuasive. Therefore, this method determines that 85% of the ETC gantry flow data is used for OD reverse thrust, and the remaining 15% is used for accuracy verification. The method of selecting the ETC gantry for accuracy verification is random sampling.

[0139] The principles for selecting the specific ETC gantries at which locations are used as the source of 85% of the reverse flow data are as follows. The four principles are all for the ETC gantries on the road network:

[0140] (1) Principle 1

[0141] In order to improve the accuracy of OD back-estimation, an ETC gantry that can cover more OD point pairs should be selected. First, determine the number of OD point pairs covered by each ETC gantry, and then select several gantries that cover a large number of OD point pairs as data sources. The OD point pairs covered by an ETC gantry means that a certain proportion of the travel demand for the OD point pair is detected through the gantry.

[0142] (2) Principle 2

[0143] To ensure that the selected ETC gantries provide independent information, avoid selecting gantries whose flow data are highly correlated with each other. Ideally, the flow of the selected gantries should provide as much independent information as possible for the OD matrix calculation. If the flow of some gantries can be derived from the flow of other gantries, then there is a certain correlation between these gantries, and these gantries should be avoided from being selected at the same time.

[0144] (3) Principle 3

[0145] For a specific OD point pair, those ETC gantries that can provide greater flow data should be selected, because greater flow usually contains more travel information. However, the same gantry may have different flow contributions for different OD point pairs, so when selecting a gantry, the gantry independence principle should also be followed.

[0146] (4) Principle 4

[0147] When selecting an ETC gantry for reverse estimation, you should ensure that the selected gantry can intercept the traffic flow to the greatest extent. Among multiple candidate gantries, you should choose the gantry combination that can capture the maximum flow. This can improve the accuracy of OD reverse estimation while ensuring rich traffic data.

[0148] Among the above four principles, Principle 2 and Principle 3 are regarded as the basic principles for locating traffic observation points.

[0149] Based on the above four principles, after the ETC gantry used for OD reverse thrust is determined within the ETC gantry range of the corresponding level, the remaining ETC gantry data are used for accuracy verification.

[0150] Real-time OD reverse prediction:

[0151] After obtaining the current expressway network cross-sectional flow and selecting 85% ETC gantries for OD inversion, the inverse process of traffic distribution (i.e., motor vehicle OD inversion technology) can be used to determine the current real-time expressway motor vehicle OD matrix. This implementation adopts the regional OD inference technology when there is no previous OD matrix.

[0152] Step 61: Set the initial values ​​of k, r, and w. The initial value of k is 1. The observed values ​​of the flow rate of each section are obtained according to the flow rate of the road network section.

[0153]

[0154] Step 62: Calculate the estimated flow value Q of each road section using the capacity-constrained iterative balance method a :

[0155] Q a =[Q a1 , Q a2 ,···,Q al ] T

[0156] in, The probability of selecting section a for the distribution from area i to area j, p i represents the generation index of zone i, A j is the attraction index of zone j, t ij is the impedance from area i to area j, where impedance is time or distance;

[0157] Construct Lagrangian function L: approximate with Lagrangian function

[0158]

[0159] make: Take the smallest variance between the actual value and the calculated value, and 5-6 is the variance calculation formula:

[0160]

[0161] Step 63: According to Q a and Update the estimate of k:

[0162] Step 64, determine whether k converges, if yes, proceed to step 64, if no, proceed to step 62;

[0163] Step 65: Calculate variance σ 2 :

[0164]

[0165] Judgment 2 Check whether the minimum value is reached. If yes, then end the process. If no, change the values ​​of r and w and go to step 1.

[0166] According to step 61 to step 65, k and r are determined to obtain q ij :

[0167]

[0168] q ij It is the element of the OD matrix, indicating the OD distribution from area i to area j.

[0169] OD reverse calculation is performed on the existing road network. If the real-time traffic data of the ETC gantry is input, the real-time OD matrix can be reversed to reflect the real-time traffic volume distribution and traffic flow trend.

[0170] The research goal of Hazhao is to conduct OD reverse calculation on the existing highway network.

[0171] Through the above-mentioned OD inverse method, by inputting the ETC traffic data obtained through AADT, the OD matrix reflecting the average conditions can be inversely deduced, which can be used as a priori OD matrix to calculate the traffic volume generated and attracted in the base year traffic zone, etc.

[0172] The OD inversion result of this embodiment is shown in OD matrix table 6.

[0173]

[0174]

[0175] Step 7. Distribute the inversely derived OD matrix to the road network to obtain the road network cross-sectional flow, and extract the road network cross-sectional flow of the section corresponding to the ETC gantry used for verification from the road network cross-sectional flow, and compare it with the actual road network cross-sectional flow of the section corresponding to the ETC gantry used for verification. If the accuracy requirement is met, end; otherwise, go to step 6 and reselect the road network cross-sectional flow of the ETC gantry used for OD inversion.

[0176] After OD back-estimation is completed through 85% of the gantry traffic data, the traffic data of the remaining ETC gantries are used for back-estimation accuracy test. Calculate the absolute value of the relative error between the actual traffic of the 15% ETC gantries used for accuracy test and the traffic redistributed to the road where the ETC is located. The OD back-estimation result is considered reliable if the average relative error is within 30%. If the back-estimation result accuracy test fails, reselect which ETC gantries are used for OD back-estimation until the back-estimation result accuracy is qualified.

[0177] The redistribution of OD inverse results on the existing road network adopts the user equilibrium allocation method, and the formula is as follows:

[0178]

[0179] The constraints are:

[0180]

[0181] x a is the traffic flow on road section a, t a is the traffic impedance on section a, t a (x a ) is the flow rate x on section a a is the traffic impedance function of the independent variable, f k rs is the flow of the kth path between ODs with the departure point r and the destination point s, If the segment a belongs to the kth path between ODs with the departure point r and the destination point s, then δ=1, otherwise δ=0;

[0182] Under the constraints, the objective function is solved and the inverse OD matrix is ​​distributed to the road network to obtain the cross-sectional flow of the road network.

[0183] In the research objectives of Hazhao, 175 of the 212 ETC gantry section traffic volumes were used to estimate the current (base year) highway motor vehicle OD matrix, and the other 37 were used to test the OD estimation accuracy. The flow distribution results, actual flow values, and relative errors of the OD inverse matrix on the 37 test sections are shown in Figure 7. The minimum relative error is 1.0%, the relative error is less than 30% accounts for 75.7%, and the average relative error is 24.7%. It can be seen from the relative error that the estimated motor vehicle OD matrix has a more reliable accuracy.

[0184] Table 7 OD matrix accuracy test

[0185]

[0186]

[0187] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. The OD inversion method based on ETC gantry flow data is characterized by: include: S1. Determine the research target as regional-level OD inversion, city-level OD inversion or highway-level OD inversion; Regional-level OD back-estimation is used to analyze traffic flows between multiple cities or regions; City-level OD back-calculation is used to analyze traffic flows within the city; Highway-level OD inversion is used to analyze traffic flow between interchanges of two-way lanes on highways; S2. Divide the component areas involved in the research target into direct traffic impact areas and indirect traffic impact areas according to the degree of impact; S3. Divide the direct traffic impact area and the indirect traffic impact area into traffic zones according to administrative units and highway sections; S4. Select ETC gantry; S5. Obtain the flow data of the selected ETC gantry in real time, and after processing, obtain the hourly traffic volume of each ETC gantry. According to the position of the ETC gantry, match the obtained hourly traffic volume to the existing road network to obtain the road network cross-sectional flow; S6, taking the traffic area in S3 as the starting and ending points, selecting the road network section flow of the ETC gantry for OD reverse calculation, and performing OD matrix reverse calculation; S7. Assign the inversely derived OD matrix to the road network to obtain the road network cross-sectional flow, and extract the road network cross-sectional flow of the section corresponding to the ETC gantry used for verification from the road network cross-sectional flow, and compare it with the actual road network cross-sectional flow of the section corresponding to the ETC gantry used for verification. If the accuracy requirement is met, end; otherwise, go to S6 and reselect the road network cross-sectional flow of the ETC gantry used for OD inversion.

2. The OD inversion method based on ETC gantry flow data according to claim 1 is characterized in that S2 include: For regional OD back-calculation, the cities and counties involved in the research target are divided into direct impact areas, and the areas outside the direct impact areas where the research target involves vehicles traveling are considered indirect impact areas; For city-level OD backcasting, all cities involved in the research target are regarded as direct impact areas, and cities adjacent to the direct impact areas are classified as indirect impact areas; For the highway-level OD back-calculation, each upstream and downstream section of the highway involved in the research target is divided into the direct impact area, and the smallest administrative unit connected to the highway entrances and exits and each interchange is regarded as the indirect impact area.

3. The OD inversion method based on ETC gantry flow data according to claim 2 is characterized in that: S3 includes: For the regional OD back-calculation, for the directly affected area, the traffic zone is divided at the district or county level, or at the township, town, or village level; for the indirectly affected area, the traffic zone is divided using an administrative unit one level larger than the directly affected area; For the city-level OD backcasting, for the directly affected areas, the traffic zones are divided according to the municipal districts, or the street-level traffic zones are divided with important traffic nodes and roads as the centroids; for the indirectly affected areas, the traffic zones are divided according to the municipal level; For highway-level OD back-calculation, for the direct impact area, each up and down section on the highway involved in the research target is divided into a traffic zone; for the indirect impact area, the districts, counties or villages directly connected to the highway entrances and exits and interchanges are divided into traffic zones.

4. The OD inversion method based on ETC gantry flow data according to claim 1 is characterized in that: The S5 includes: Obtain the 15-minute traffic data of the selected ETC gantry in real time, convert the traffic of different models into PCU traffic through the vehicle conversion coefficient, sum up to get the PCU flow rate of each ETC gantry, and process it into the hourly traffic volume of each ETC gantry; Match the location of the ETC gantry with the road network, assign the corresponding hourly flow data to the corresponding ETC gantry, and obtain the current road network cross-sectional flow.

5. The OD inversion method based on ETC gantry flow data according to claim 1 is characterized in that: The S5 includes: S51, obtaining the selected ETC gantry traffic data, the data time range is one year; S52. For each ETC gantry, convert the flow data of each vehicle type into flow in pcu according to the respective vehicle conversion coefficients, and sum the pcu flows of each vehicle type to obtain the total flow in pcu of each gantry; S53, for each gantry, the total traffic volume is divided by 365 days to obtain the annual average daily traffic volume of each ETC gantry; S55. For each ETC gantry, the hourly traffic volume of each gantry is obtained based on the annual average daily traffic volume and the peak hour ratio; S56. Match the location of the ETC gantry with the road network, assign the corresponding hourly flow data to the corresponding ETC gantry, and obtain the current road network cross-sectional flow.

6. The OD inversion method based on ETC gantry flow data according to claim 1 is characterized in that S6 include: S61, set the initial values ​​of k, r, and w, where the initial value of k is 1, and the observed values ​​of the flow of each section are obtained according to the flow of the road network section. S62, using the capacity-limited iterative balance method to calculate the estimated flow value Q of each road section a : Q a =[Q a1 ,Q a2 ,···,Q al ] T in, The probability of selecting section a for the distribution from area i to area j, p i represents the generation index of zone i, A j is the attraction index of zone j, t ij is the impedance from area i to area j, where impedance is time or distance; S63, according to Q a and Update the estimate of k: S64, judging whether k converges, if so, proceeding to S64, if not, proceeding to S62; S65. Calculate variance σ 2 : Judgment 2 Has it reached the minimum? If so, end. If not, change the values ​​of r and w and go to S1. According to S61 to S65, the determined k, r get q ij : q ij It is the element of the OD matrix, indicating the OD distribution from area i to area j.

7. The OD inversion method based on ETC gantry flow data according to claim 2 is characterized in that S4 include: For regional OD back-propagation, all ETC gantries in the direct impact area are considered as candidates, and some ETC gantries in the indirect impact area are included as candidates; For city-level OD backpropagation, all ETC gantries in the direct impact area will be considered as alternatives; For highway-level OD backpropagation, all ETC gantries in the traffic area are considered as candidates.

8. The OD inverse estimation method based on ETC gantry flow data according to claim 7 is characterized in that: In S6, the road network cross-sectional flow of the ETC gantry selected for OD reverse estimation includes: According to the following principles, the network cross-sectional flow of the ETC gantries among the candidates is selected for OD reverse estimation, and the network cross-sectional flow of the remaining ETC gantries is used for verification; Principle 1: Determine the number of OD point pairs covered by each ETC gantry, and then select several gantries that cover a large number of OD point pairs; Principle 2: Avoid selecting ETC gantries with highly correlated traffic data; Principle 3: For a specific OD point pair, select the gantry that can capture the maximum flow. Principle 4: Among multiple alternative gantries, those gantry combinations that can capture the maximum flow should be selected.

9. The OD inversion method based on ETC gantry flow data according to claim 8 is characterized in that: The method to distribute the inverse OD matrix to the road network is: Establish the objective function: The constraints are: Where: x a is the traffic flow on section a, t a is the traffic impedance on section a, t a (x a ) is the flow rate x on section a a is the traffic impedance function of the independent variable, f k rs is the flow of the kth path between ODs with the departure point r and the destination point s, If the segment a belongs to the kth path between ODs with the departure point r and the destination point s, then δ=1, otherwise δ=0; Under the constraints, the objective function is solved and the inverse OD matrix is ​​distributed to the road network to obtain the cross-sectional flow of the road network.

10. An OD reverse estimation device based on ETC gantry flow data, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the OD inversion method based on ETC gantry traffic data as described in any one of claims 1 to 7.

Images