Non-uniform discretization method of road network for single-center isochrone generation

Abstract

The application discloses a road network uneven discrete method for single-center traffic isochrone generation, comprising the following steps: 1) selecting a studied road network, and obtaining road network topological structure basic data; 2) analyzing spatial autocorrelation variation of data sets under different discrete densities and estimation value error of same observation point positions, and determining a reasonable road network average discrete density range; 3) determining plot discrete density under the condition of being lower than the road network average discrete density range; 4) combining road network area, and calculating target total discrete point number; 5) establishing a key node identification index system, and determining key nodes and key road sections based on an entropy weight method and grey correlation degree sorting; 6) calculating actual surface discrete point number, node discrete point number and road section discrete point number, and adding to obtain actual total discrete point number; and 7) judging error of the actual total discrete point number and the target total discrete point number, and outputting a road network uneven discrete scheme. The application can reduce calculation amount and improve accuracy of traffic isochrone generation.

Description

Technical Field

[0001] This invention relates to the technical field of time-based performance evaluation and analysis of urban road networks, and in particular to a non-uniform discretization method for road networks that considers the distribution characteristics of road network geospatial information in traffic isochrone generation algorithms. Background Technology

[0002] The timeliness performance of a road network reflects its capacity to handle traffic flow over time and is a crucial evaluation parameter for road network planning, design, and management. Traffic isochrones are closed curves formed by connecting all points in the road network with equal travel times, reflecting the spatial distribution of travel times within the network. They are a powerful tool for evaluating the spatial timeliness performance of road networks. Accurately generating discrete points in locations rich in geospatial information is fundamental to generating traffic isochrones. Existing traffic isochrone generation algorithms neglect the uneven distribution of spatial information within the road network, often generating discrete points uniformly.

[0003] Existing uniform discretization methods with no indifference have two shortcomings: 1. They ignore the differences in importance caused by road and traffic attributes of road segments and intersections. This importance indicates that these road segments and intersections are more representative and contain richer spatial information; 2. Higher road network discretization uniformity density generates more discrete sampling points, resulting in a larger amount of total spatial information and higher estimation accuracy. However, this improvement in accuracy has a saturation point, and blindly increasing the discretization density will increase unnecessary computation. This invention, based on the characteristics of uneven geographic spatial distribution, proposes a non-uniform road network discretization rule while considering both computational load and estimation error, thus determining the non-uniform road network discretization method.

[0004] This invention, when acquiring discrete points in a road network, generates a time-domain representation of the urban road network using sampling datasets of different discrete densities. It analyzes the relationship between the road network's discrete density range and estimation error to determine a reasonable average discrete density range for the road network considering the saturation of spatial information growth. It identifies high-discrete-density locations containing rich spatial information, namely key road network segments. A key segment identification index system composed of road network and traffic indicators is established. Based on the entropy weight method and grey relational analysis, high-discrete-density locations containing rich spatial information are identified, thereby enabling non-uniform discretization of the road network. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a non-uniform discretization method for generating isochronous traffic lines in a single center. This method considers the reasonable average discretization density of spatial information and determines the location of high discretization density, breaking through the existing uniform discretization method that ignores the distribution characteristics of spatial information. It improves the accuracy of estimation results while reducing unnecessary computation, and provides a basis for isochronous eigenvalue estimation.

[0006] To achieve the above objectives, the technical solution provided by this invention is: a method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, comprising the following steps:

[0007] 1) Select the road network to be studied and obtain basic data on the road network topology;

[0008] 2) On the selected road network, determine the center point of the traffic isochrones and the discrete density gradient; obtain the travel time data of high-density discrete points as the basic sampling dataset; and determine a reasonable range of average discrete density of the road network by analyzing the changes in spatial autocorrelation of the dataset under different discrete densities and the estimation error of the same observation point location.

[0009] 3) Under the condition of being below the average discrete density range of the road network, the discrete densities of different plots are divided into different experimental groups according to the gradient, and hypothesis testing is performed. The average standard error and significance P value of each group in the T test are compared to determine the discrete density of the plots.

[0010] 4) Based on the determined average discrete density of the road network and the discrete density of the land parcels, and in conjunction with the road network area, calculate the total number of target discrete points.

[0011] 5) After determining the total number of target discrete points, select a key node identification index system that includes road network indicators and traffic indicators, calculate the index values ​​and sort the correlation of each node and road segment, determine the number of key nodes and key road segments based on a specific ratio, and determine the key nodes and key road segments according to the correlation sort.

[0012] 6) After obtaining the key nodes and key road segments, calculate the actual number of surface discrete points, node discrete points, and road segment discrete points respectively, and add them together to obtain the total number of actual discrete points; wherein, the surface discrete points are the surface discrete points of the land parcel generated by the geometric center of the grid when the land parcel is uniformly divided into a series of grids with a specific density; the node discrete points are the discrete points at the intersections of the road network; and the road segment discrete points are the discrete points generated on the road segments corresponding to the edges connecting the nodes.

[0013] 7) Determine whether the difference between the actual total number of discrete points and the target total number of discrete points is within the error range. If it is within the error range, output the uneven road network discrete scheme. If it is not within the error range, return to the part in the previous step that determined the number of discrete points of the road segment, readjust the discrete density of the key road segment, change the number of discrete points of the road segment, and output the discrete scheme after the actual total number of discrete points reaches the requirement.

[0014] Furthermore, in step 1), the basic data of the road network topology includes the road network area, nodes in the road network, road segments and their connection relationships, road grade of each road segment, current road conditions, current intersection conditions and current signal timing.

[0015] Furthermore, in step 2), determining a reasonable range for the average discrete density of the road network includes the following steps:

[0016] 2.1) Determine the discrete density gradient and traffic isochrone center points of the target road network. The discrete density gradient is the gradient sampled from N discrete points at different proportions. It is assumed that the traffic isochrones generated from M discrete points reflect the actual isochronous state; therefore, the set of M sampled points constitutes the basic sampling dataset. The discrete density gradient proportions are selected as 0.1. 2 M discrete points, 0.2 2 M discrete points, ..., 0.9 2 M discrete points;

[0017] 2.2) Generate high-density discrete points. Use online map platforms and measured data to obtain discrete point travel time network data at a certain time as the basic sampling dataset for the study, which represents the actual observation value. Based on the idea of ​​spatial sampling, sample from the basic sampling dataset according to the gradient at a specific ratio to form sampling datasets with different discrete densities. Randomly generate discrete points with different discrete densities within the research range of the target road network.

[0018] 2.3) Spatial Autocorrelation Analysis of Datasets with Different Discrete Densities: Since isochronous eigenvalues ​​are spatially continuous, they should exhibit significant spatial autocorrelation. To ensure that the generated traffic isochrones reflect the spatial distribution characteristics of isochronous eigenvalues ​​at a given moment, the discrete density sampling dataset must pass a significance test for spatial autocorrelation. The global Moran's index (MI) is used to evaluate the spatial autocorrelation of different discrete density sampling datasets. For hypothesis testing analysis of spatial autocorrelation, firstly, the Z-test score and p-value are used to determine whether the hypothesis is significant. Then, the MI value is used to determine whether the data exhibits spatial autocorrelation. The MI value is distributed in [-1, 1]. When the value is greater than 0, it indicates a positive correlation; when the value is less than 0, it indicates a negative correlation; and when the value is equal to 0, it indicates no correlation.

[0019] 2.4) Estimation Error Analysis: Ordinary kriging interpolation is used to generate isochronous data. The error between the estimated values ​​and actual observed values ​​at the same observation points for isochronous data with different discrete densities is calculated. This simplifies the calculation of the estimation error of traffic isochronous lines to the calculation of the estimation error of the interpolated isochronous data. The high discrete density base sampling dataset from step 2.2) is used as the benchmark value of the estimation error. The root mean square error (RMSE) and mean absolute error rate (MAER) are used to evaluate the magnitude of the estimation error, and the coefficient of variation (CV) is used to evaluate the degree of dispersion of the estimation error.

[0020]

[0021]

[0022]

[0023] In the formula: —Estimated isochronous eigenvalues ​​of the i-th observation point in the isochronous domain generated by different discrete density sampling datasets;

[0024] —The actual observed isochronous characteristic value of the i-th observation point;

[0025] —The average of the estimated isochronous eigenvalues ​​for observation points with different discrete densities;

[0026] n — the number of observation points;

[0027] k—the sampling proportion of the sampled dataset in the base sampled dataset. When k equals 0.1, it means that 0.1 points are drawn from the M discrete points in the base sampled dataset. 2 M discrete points;

[0028] 2.5) By observing the changes in travel time (MI) values ​​of different discrete gradients of the road network and the relationship between discrete density and error index and coefficient of variation, a reasonable range of discrete density is selected, which meets the following requirements: ① The isochronous characteristic value Z test is significant within this density range; ② Under the condition that it is less than or equal to this density range, as the discrete density increases, the MI value increases significantly and the estimation error and its coefficient of variation decrease significantly.

[0029] Furthermore, in step 3), a paired-samples t-test is performed on the estimation results of the sampled dataset. Since the spatial information carried by discrete points on roads in the road network is richer than that carried by discrete points on land parcels, the discrete rule is set as the discrete density of land parcels being less than the average discrete density of the road network. Therefore, the study only performs hypothesis testing on the density of uniformly distributed discrete points below the average discrete density of the road network. The null hypothesis is that there is a significant difference between the estimated and observed data. If the p-value is less than 0.1, the null hypothesis is accepted, indicating that there is a significant difference between the two data sets, that is, the estimated value cannot correctly reflect the observed value. If the p-value is greater than 0.1, the null hypothesis is rejected, indicating that the estimated value is consistent with the observed value. The smaller the average standard error, the larger the p-value, indicating that the estimation effect is better. As the discrete density of land parcels increases, the average standard error of the paired samples gradually decreases, and the significance p-value gradually increases. At a certain density, both the significance p-value and the average standard error tend to stabilize, and this density is selected as the discrete density of the land parcels.

[0030] Furthermore, in step 4), the total number of target discrete points N0 is determined, and the calculation formula is as follows:

[0031]

[0032] In the formula, A networkFor the target road network area, Let N0 be the average discrete density of the target road network, where N0 is an integer.

[0033] Furthermore, in step 5), key nodes and key road segments are identified, including the following steps:

[0034] 5.1) Taking into account the topological characteristics and traffic operation characteristics of the road network, and referring to the existing research results on the evaluation indicators of key nodes of the road network, a key node identification indicator system is established from the two aspects of road network indicators and traffic indicators.

[0035] 5.2) The method for determining the weights of key nodes and key road sections based on the entropy weight method is as follows:

[0036] Step 1: Establish an initial matrix X, which contains n′ nodes and k′ evaluation indicators, where x fg This represents the data for the g-th evaluation index of the f-th node;

[0037] X = (x fg ) n′×k′ , f=1,2,…,n′, g=1,2,…,k′

[0038] Step 2: Normalize the initial matrix X to obtain the index normalized matrix B, x min and x max Indicates the maximum and minimum values ​​under the same evaluation index;

[0039]

[0040] B = (b fg ) n′×k′

[0041] In the formula, b fg This represents the element in the f-th row and g-th column of the normalized matrix B;

[0042] Step 3: Calculate the entropy S of the g-th evaluation index. g ;

[0043]

[0044]

[0045] In the formula, P fg The weight of the index value of the f-th node under the g-th evaluation index;

[0046] Step 4: Calculate the entropy weight ω of the g-th evaluation index. g ;

[0047]

[0048] 5.3) Ranking of Nodes Based on Grey Relational Analysis: Grey relational degree is the degree of similarity or dissimilarity in the development trends of factors in a system. The higher the grey relational degree, the higher the homogeneity of the change trends between the two factors, that is, the higher the degree of correlation between them. The maximum value of each evaluation index is taken to construct a reference sequence. The higher the grey relational degree between the index sequence of other nodes and the reference sequence, the higher the importance of the node. The average relational degree of the nodes connected to the road segment is calculated as the basis for ranking the importance of the road segment. The steps are as follows:

[0049] Step 1: Dimensionless processing of raw indicator data: The result of the mean method can best reflect the differences in the degree of variation of the raw data of each indicator. Therefore, the mean method is used to process the raw data into dimensionless form.

[0050]

[0051] In the formula, This refers to the data after dimensionless processing of the data for the g-th evaluation index at the f-th node;

[0052] Step 2: Construct a reference sequence Y for evaluating node importance. * ;

[0053]

[0054] In the formula, Let be the maximum value of the k′-th evaluation index at the f-th node after dimensionless processing, denoted as

[0055] Step 3: Calculate the correlation coefficient r of each evaluation indicator. f (g);

[0056]

[0057]

[0058] In the formula, A is the reference sequence Y evaluated by the importance of nodes. * and dimensionless data The constructed matrix This represents the data after dimensionless processing of the data for the k′-th evaluation index at the n′-th node, r. f (g) represents the correlation coefficient of the g-th evaluation index at the f-th intersection, Δ min and Δ max denoted as the maximum and minimum values ​​in matrix A, and ρ is the resolution coefficient;

[0059] Step 4: Calculate the correlation degree R(f) of each node. The correlation degree is the importance of the node. The higher the correlation degree, the more important the node is.

[0060]

[0061] Step 5: Determine the correlation degree of all road segments based on the adjacency matrix;

[0062]

[0063] In the formula, R(ab) represents the road segment e connecting nodes a and b. ab The degree of correlation;

[0064] 5.4) Determine the number of key nodes and key road segments based on a specific ratio, and rank them according to their correlation. For a road network with n” nodes and m road segments, determine the ratio p’ of key nodes and key road segments, then the number of key nodes n key =round(n”p′,0), where m is the number of critical road segments. key =round(mp′,0); Sort the node importance R(f) in ascending order to form a node importance sequence R. node ={R(f) min ,...,R(f) max}, the first n of the sequence key The nodes corresponding to each degree of correlation are the key nodes of the road network when the ratio is p′, forming the key node set V. key ∈V, where V is the set of all nodes in the road network. The road segment importance sequence R is formed by sorting the road segments by their correlation degree R(ab) from largest to smallest. edge ={R(ab)} min ,...,R(ab) max}, the first m of the sequence key The road segments corresponding to each degree of correlation are the key road segments of the road network when the proportion is p′, and the set E of key road segments is formed by this. key ∈E, where E is the set of all road segments in the road network.

[0065] Furthermore, in step 6), the actual total number of discrete points is calculated, including the following steps:

[0066] a. Determine the number of discrete points on the surface: The land parcel is uniformly divided into a series of grids with a specific density. The geometric center of the grid generates discrete points on the surface of the land parcel. However, some discrete points on the surface also contain discrete points on the road lines. These are duplicate discrete points on the surface and will be deleted when calculating isochronous eigenvalues. First, determine the repetition rate of the discrete points on the surface. Then, calculate the number of discrete points on the surface using the following formula:

[0067]

[0068] In the formula, k is an integer representing the number of discrete points on the surface. min The minimum discrete density of the land surface to ensure that isochronous interpolation does not produce holes;

[0069] b. Determine the number of discrete nodes: key nodes The corresponding number of discrete points is Non-critical node v f The number of discrete points is 1, and the total number of discrete points in the node is N. V Its value is equal to the sum of the discrete points of all critical nodes and non-critical nodes; where, To be with key nodes The number of connected road segments;

[0070] c. Determine the number of discrete points for each road segment: Divide the road segments into two categories: critical road segments and non-critical road segments. First, based on the road segment correlation score from step 5.3), determine the difference coefficient μ between the two categories of road segments. r It is the ratio of the discrete density of critical road segments to the discrete density of non-critical road segments, and is equal to the ratio of the average correlation degree of all road segments in the road segment set; then, the discrete density of critical road segments and non-critical road segments are calculated separately; finally, the number of discrete points of all actual road segments is calculated based on the discrete density. Their sum is the number of discrete points in the road segment. The calculation formula is as follows:

[0071]

[0072]

[0073]

[0074] In the formula, k is an integer; key Discrete density of critical road segments; l ab For road segment e ab Length; E key A collection of key road sections;

[0075] d. Calculate the actual total number of discrete points: After adjusting the discrete points according to the discretization rules, the actual total number of discrete points N′ is the sum of the adjusted surface discrete points, node discrete points, and road segment discrete points;

[0076]

[0077] Furthermore, in step 7), it is determined whether the difference between the actual total number of discrete points and the target total number of discrete points is within the error range. If it is not within the error range, the process returns to step c to readjust the discrete density k of the critical road segment. keyIf within the error range, output the discrete density of critical road segments, kkey, and the discrete density of non-critical road segments, k″, and output the non-uniform discrete road network scheme.

[0078] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0079] 1) It is designed for single-center traffic isochrone generation, while also generating discrete points on road segments and plots to improve the accuracy of traffic isochrones.

[0080] 2) By discretizing the road network unevenly, the accuracy of the estimation results can be improved while reducing unnecessary computation.

[0081] 3) Based on the directed complex network, a road network topology is established. Taking into account the traffic attributes and topology of the road network, the intersections and road segments with high importance of the road network are identified, and the discrete density of these key locations is increased, so that the high discrete density locations contain richer spatial information and the generated isochrones are more accurate.

[0082] In summary, this invention can improve the accuracy of traffic isochrone generation while reducing computational load. It can be applied to both existing and planned road networks, has practical application value, and is worthy of promotion. Attached Figure Description

[0083] Figure 1 This is a flowchart of the method of the present invention.

[0084] Figure 2 A flowchart for determining the reasonable average discrete density range of a road network.

[0085] Figure 3 This is a flowchart of the key point segment identification method.

[0086] Figure 4 This is a flowchart of a non-uniform discrete road network.

[0087] Figure 5 This is a road network diagram for the case study.

[0088] Figure 6 This is a diagram showing the results of the uneven discrete road network in the case study. Detailed Implementation

[0089] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0090] like Figure 1 As shown, this embodiment discloses a method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, including the following steps:

[0091] 1) Select the road network to be studied and obtain basic data on the road network topology; the basic data on the road network topology includes the road network area, nodes in the road network, road segments and their connection relationships, road grade of each road segment, current road conditions, current intersection conditions and current signal timing.

[0092] 2) On the selected road network, determine the center point of the traffic isochrones and the discrete density gradient; obtain travel time data of high-density discrete points as the basic sampling dataset; by analyzing the spatial autocorrelation changes of the dataset under different discrete densities and the estimation error of the same observation point location, determine a reasonable range of average discrete density of the road network, including the following steps:

[0093] 2.1) Determine the discrete density gradient and traffic isochrone center points of the target road network. The discrete density gradient is the gradient sampled from N discrete points at different proportions. It is assumed that the traffic isochrones generated from M discrete points reflect the actual isochronous state; therefore, the set of M sampled points constitutes the basic sampling dataset. The discrete density gradient proportions are selected as 0.12M discrete points and 0.2... 2 M discrete points, ..., 0.9 2 M discrete points;

[0094] 2.2) Generate high-density discrete points. Use online map platforms and measured data to obtain discrete point travel time network data at a certain time as the basic sampling dataset for the study, which represents the actual observation value. Based on the idea of ​​spatial sampling, sample from the basic sampling dataset according to the gradient at a specific ratio to form sampling datasets with different discrete densities. Randomly generate discrete points with different discrete densities within the research range of the target road network.

[0095] 2.3) Spatial Autocorrelation Analysis of Datasets with Different Discrete Densities: Since isochronous eigenvalues ​​are spatially continuous, they should exhibit significant spatial autocorrelation. To ensure that the generated traffic isochrones reflect the spatial distribution characteristics of isochronous eigenvalues ​​at a given moment, the discrete density sampling dataset must pass a significance test for spatial autocorrelation. The global Moran's index (MI) is used to evaluate the spatial autocorrelation of different discrete density sampling datasets. For hypothesis testing analysis of spatial autocorrelation, firstly, the Z-test score and p-value are used to determine whether the hypothesis is significant. Then, the MI value is used to determine whether the data exhibits spatial autocorrelation. The MI value is distributed in [-1, 1]. When the value is greater than 0, it indicates a positive correlation; when the value is less than 0, it indicates a negative correlation; and when the value is equal to 0, it indicates no correlation.

[0096] 2.4) Estimation Error Analysis: Ordinary kriging interpolation is used to generate isochronous data. The error between the estimated values ​​and actual observed values ​​at the same observation points for isochronous data with different discrete densities is calculated. This simplifies the calculation of the estimation error of traffic isochronous lines to the calculation of the estimation error of the interpolated isochronous data. The high discrete density base sampling dataset from step 2.2) is used as the benchmark value of the estimation error. The root mean square error (RMSE) and mean absolute error rate (MAER) are used to evaluate the magnitude of the estimation error, and the coefficient of variation (CV) is used to evaluate the degree of dispersion of the estimation error.

[0097]

[0098]

[0099]

[0100] In the formula: —Estimated isochronous eigenvalues ​​of the i-th observation point in the isochronous domain generated by different discrete density sampling datasets;

[0101] —The actual observed isochronous characteristic value of the i-th observation point;

[0102] —The average of the estimated isochronous eigenvalues ​​for observation points with different discrete densities;

[0103] n — the number of observation points;

[0104] k — the sampling ratio of the sampling dataset to the base dataset. When k equals 0.1, it means that 0.12M discrete points are drawn from the M discrete points in the base sampling dataset.

[0105] 2.5) By observing the changes in travel time (MI) values ​​of different discrete gradients of the road network and the relationship between discrete density and error index and coefficient of variation, a reasonable range of discrete density is selected, which meets the following requirements: ① The isochronous characteristic value Z test is significant within this density range; ② Under the condition that it is less than or equal to this density range, as the discrete density increases, the MI value increases significantly and the estimation error and its coefficient of variation decrease significantly.

[0106] 3) Under conditions below the average discrete density range of the road network, the discrete densities of different land parcels are divided into different experimental groups according to the gradient, and hypothesis testing is performed. The mean standard error and significance p-value of each group in the T-test are compared to determine the discrete density of the land parcels, as follows:

[0107] Paired-samples t-tests were performed on the estimation results of the sampled dataset. Since discrete points on roads in the road network carry richer spatial information than discrete points on land parcels, the dispersion rule was set as the discrete density of land parcels being less than the average discrete density of the road network. Therefore, the study only performed hypothesis testing on the density of uniformly distributed discrete points below the average dispersion of the road network. The null hypothesis was that there was a significant difference between the estimated and observed data. If the p-value was less than 0.1, the null hypothesis was accepted, indicating a significant difference between the two sets of data, meaning that the estimated value could not accurately reflect the observed value. If the p-value was greater than 0.1, the null hypothesis was rejected, indicating that the estimated value was consistent with the observed value. The smaller the mean standard error, the larger the p-value, indicating a better estimation effect. As the discrete density of land parcels increased, the mean standard error of the paired samples gradually decreased, and the significance p-value gradually increased. At a certain density, both the significance p-value and the mean standard error tended to stabilize, and this density was selected as the discrete density of the land parcels.

[0108] 4) Based on the determined average discrete density of the road network and the discrete density of land parcels, and in conjunction with the road network area, calculate the total number of target discrete points N0. The calculation formula is as follows:

[0109]

[0110] In the formula, A network For the target road network area, Let N0 be the average discrete density of the target road network, where N0 is an integer.

[0111] 5) After determining the total number of target discrete points, select a key node identification index system that includes road network indicators and traffic indicators. Calculate the index values ​​and rank the correlation of each node and road segment. Determine the number of key nodes and key road segments based on a specific ratio. The key nodes and key road segments are determined according to the correlation ranking, including the following steps:

[0112] 5.1) Taking into account the topological characteristics and traffic operation characteristics of the road network, and referring to the existing research results on the evaluation indicators of key nodes of the road network, a key node identification indicator system is established from the two aspects of road network indicators and traffic indicators.

[0113] 5.2) The method for determining the weights of key nodes and key road sections based on the entropy weight method is as follows:

[0114] Step 1: Establish an initial matrix X, which contains n′ nodes and k′ evaluation indicators, where x fg This represents the data for the g-th evaluation index of the f-th node;

[0115] X = (x fg ) n′×k′ , f=1,2,…,n′, g=1,2,…,k′

[0116] Step 2: Normalize the initial matrix X to obtain the index normalized matrix B, x min and x max Indicates the maximum and minimum values ​​under the same evaluation index;

[0117]

[0118] B = (b fg ) n′×k′

[0119] In the formula, b fg This represents the element in the f-th row and g-th column of the normalized matrix B;

[0120] Step 3: Calculate the entropy S of the g-th evaluation index. g ;

[0121]

[0122]

[0123] In the formula, P fg The weight of the index value of the f-th node under the g-th evaluation index;

[0124] Step 4: Calculate the entropy weight ω of the g-th evaluation index. g ;

[0125]

[0126] 5.3) Ranking of Nodes Based on Grey Relational Analysis: Grey relational degree is the degree of similarity or dissimilarity in the development trends of factors in a system. The higher the grey relational degree, the higher the homogeneity of the change trends between the two factors, that is, the higher the degree of correlation between them. The maximum value of each evaluation index is taken to construct a reference sequence. The higher the grey relational degree between the index sequence of other nodes and the reference sequence, the higher the importance of the node. The average relational degree of the nodes connected to the road segment is calculated as the basis for ranking the importance of the road segment. The steps are as follows:

[0127] Step 1: Dimensionless processing of raw indicator data: The result of the mean method can best reflect the differences in the degree of variation of the raw data of each indicator. Therefore, the mean method is used to process the raw data into dimensionless form.

[0128]

[0129] In the formula, This refers to the data after dimensionless processing of the data for the g-th evaluation index at the f-th node;

[0130] Step 2: Construct a reference sequence Y for evaluating node importance. * ;

[0131]

[0132] In the formula, Let be the maximum value of the k′-th evaluation index at the f-th node after dimensionless processing, denoted as

[0133] Step 3: Calculate the correlation coefficient r of each evaluation indicator. f (g);

[0134]

[0135]

[0136] In the formula, A is the reference sequence Y* used for evaluating node importance and the dimensionless processed data. The constructed matrix This represents the data after dimensionless processing of the data for the k′-th evaluation index at the n′-th node, r. f (g) represents the correlation coefficient of the g-th evaluation index at the f-th intersection, Δ min and Δ max denoted as the maximum and minimum values ​​in matrix A, and ρ is the resolution coefficient;

[0137] Step 4: Calculate the correlation degree R(f) of each node. The correlation degree is the importance of the node. The higher the correlation degree, the more important the node is.

[0138]

[0139] Step 5: Determine the correlation degree of all road segments based on the adjacency matrix;

[0140]

[0141] In the formula, R(ab) represents the road segment e connecting nodes a and b. ab The degree of correlation;

[0142] 5.4) Determine the number of key nodes and key road segments based on a specific ratio, and rank them according to their correlation. For a road network with n” nodes and m road segments, determine the ratio p’ of key nodes and key road segments, then the number of key nodes n key =round(n”p′,0), where m is the number of critical road segments. key =round(mp′,0); Sort the node importance R(f) in ascending order to form a node importance sequence R. node ={R(f) min ,...,R(f) max}, the first n of the sequencekey The nodes corresponding to each degree of correlation are the key nodes of the road network when the ratio is p′, forming the key node set V. key ∈V, where V is the set of all nodes in the road network. The road segment importance sequence R is formed by sorting the road segments by their correlation degree R(ab) from largest to smallest. edge ={R(ab)} min ,...,R(ab) max}, the first m of the sequence key The road segments corresponding to each degree of correlation are the key road segments of the road network when the proportion is p′, and the set E of key road segments is formed by this. key ∈E, where E is the set of all road segments in the road network.

[0143] 6) After obtaining the key nodes and key road segments, calculate the actual number of surface discrete points, node discrete points, and road segment discrete points respectively, and add them together to obtain the total number of actual discrete points; wherein, the surface discrete points are the surface discrete points of the land parcel generated by the geometric center of the grid when the land parcel is uniformly divided into a series of grids with a specific density; the node discrete points are the discrete points at the intersections of the road network; and the road segment discrete points are the discrete points generated on the road segments corresponding to the edges connecting the nodes.

[0144] Calculating the actual total number of discrete points includes the following steps:

[0145] a. Determine the number of discrete points on the surface: The land parcel is uniformly divided into a series of grids with a specific density. The geometric center of the grid generates discrete points on the surface of the land parcel. However, some discrete points on the surface also contain discrete points on the road lines. These are duplicate discrete points on the surface and will be deleted when calculating isochronous eigenvalues. First, determine the repetition rate of the discrete points on the surface. Then, calculate the number of discrete points on the surface using the following formula:

[0146]

[0147] In the formula, k is an integer representing the number of discrete points on the surface. min The minimum discrete density of the land surface to ensure that isochronous interpolation does not produce holes;

[0148] b. Determine the number of discrete nodes: key nodes The corresponding number of discrete points is Non-critical node v f The number of discrete points is 1, and the total number of discrete points in the node is N. V Its value is equal to the sum of the discrete points of all critical nodes and non-critical nodes; where, To be with key nodes The number of connected road segments;

[0149] c. Determine the number of discrete points for each road segment: Divide the road segments into two categories: critical road segments and non-critical road segments. First, based on the road segment correlation score from step 5.3), determine the difference coefficient μ between the two categories of road segments. r It is the ratio of the discrete density of critical road segments to the discrete density of non-critical road segments, and is equal to the ratio of the average correlation degree of all road segments in the road segment set; then, the discrete density of critical road segments and non-critical road segments are calculated separately; finally, the number of discrete points of all actual road segments is calculated based on the discrete density. Their sum is the number of discrete points in the road segment. The calculation formula is as follows:

[0150]

[0151]

[0152]

[0153] In the formula, k is an integer; key Discrete density of critical road segments; l ab For road segment e ab Length; E key A collection of key road sections;

[0154] d. Calculate the actual total number of discrete points: After adjusting the discrete points according to the discretization rules, the actual total number of discrete points N′ is the sum of the adjusted surface discrete points, node discrete points, and road segment discrete points:

[0155]

[0156] 7) Determine if the difference between the actual total number of discrete points and the target total number of discrete points is within the error range. If it is not within the error range, return to step c to readjust the discrete density k of the critical road segment. key The number of discrete points for each road segment is adjusted until the total number of discrete points reaches the required level, at which point the discrete solution is output. If the result is within the error range, the discrete density k of the critical road segments is output. key Given the discrete density k″ of non-critical road segments, output the non-uniform discrete scheme of the road network.

[0157] Below, we select the central business district of Tianhe District as the road network area, which is 8.56 km². 2 The key study area of ​​the road network extends north to Tianhe Road, south to Linjiang Avenue, west to Guangzhou Avenue Central, and east to Liede Avenue. Specific boundaries are as follows: Figure 5 As shown, the road network comprises 140 road sections, of which expressways, arterial roads, secondary arterial roads, and local roads account for 27.2%, 5.7%, 35.7%, and 31.4%, respectively.

[0158] according to Figure 3The method shown identifies 17 key nodes and 45 key road segments in the case study road network. Based on... Figure 2 The method shown determines the area to be 8.56 km². 2 The reasonable average discrete density of the case road network is 190 discrete points / square kilometer, and the discrete density of the plots is 105 discrete points / square kilometer. The following section will calculate the number of discrete points for different categories of the case road network based on the non-uniform road network discretization method and output the non-uniform road network discretization scheme, such as... Figure 4 As shown:

[0159] 1) For ease of calculation, the total number of discrete points N0 is taken as 1600;

[0160] 2) There are 68 critical nodes and 79 ordinary nodes, with a total of N discrete nodes. V =147;

[0161] 3) Simple calculations show that, assuming all discrete points are retained, the road segment discrete point density is lowest. The distances between discrete points in critical and non-critical road segments are 61.40m and 69.94m, respectively, both less than the grid side length of 97.52m. Therefore, all discrete points corresponding to the plot grids spatially associated with road segments are Class 1 points, and the calculated discrete point repetition rate... Number of discrete points on the surface

[0162] 4) The ratio of the mean correlation between critical road segments and non-critical road segments is the coefficient of variation, μ. r =0.878, then the discrete density k of the critical road segment is... key = 27.09 discrete points / km, and the discrete density of non-critical road sections k″ = 23.78 discrete points / km;

[0163] 5) After adjustment, there are 339 key road segment discrete points and 579 non-key road segment discrete points that conform to the road network non-uniform discrete rules. At this time, the actual total number of discrete points N′=1610.

[0164] The results of the non-uniform discretization of the road network in the case are as follows: Figure 6 As shown.

[0165] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, characterized in that, Includes the following steps: 1) Select the road network to be studied and obtain basic data on the road network topology; 2) On the selected road network, determine the center point of the traffic isochrones and the discrete density gradient; obtain the travel time data of high-density discrete points as the basic sampling dataset; and determine a reasonable range of average discrete density of the road network by analyzing the changes in spatial autocorrelation of the dataset under different discrete densities and the estimation error of the same observation point location. 3) Under the condition of being below the average discrete density range of the road network, the discrete densities of different plots are divided into different experimental groups according to the gradient, and hypothesis testing is performed. The average standard error and significance P value of each group in the T test are compared to determine the discrete density of the plots. 4) Based on the determined average discrete density of the road network and the road network area, calculate the total number of target discrete points; 5) After determining the total number of target discrete points, select a key node identification index system that includes road network indicators and traffic indicators, calculate the index values ​​and sort the correlation of each node and road segment, determine the number of key nodes and key road segments based on a specific ratio, and determine the key nodes and key road segments according to the correlation sort. 6) After obtaining the key nodes and key road segments, calculate the actual number of surface discrete points, node discrete points, and road segment discrete points respectively, and add them together to obtain the total number of actual discrete points. Among them, surface discrete points are generated by uniformly dividing the land parcel into a series of grids at a specific density, and the geometric center of the grid is the generated land parcel surface discrete point; node discrete points are discrete points at road network intersections; road segment discrete points are discrete points generated on the road segments corresponding to the edges connecting nodes. 7) Determine whether the difference between the actual total number of discrete points and the target total number of discrete points is within the error range. If it is within the error range, output the uneven road network discrete scheme. If it is not within the error range, return to the part in the previous step that determined the number of discrete points of the road segment, readjust the discrete density of the key road segment, change the number of discrete points of the road segment, and output the discrete scheme after the actual total number of discrete points reaches the requirement.

2. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 1, is characterized in that: In step 1), the basic data of the road network topology includes the road network area, nodes in the road network, road segments and their connection relationships, road grade of each road segment, current road conditions, current intersection conditions and current signal timing.

3. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 2, is characterized in that: In step 2), determining a reasonable range for the average discrete density of the road network includes the following steps: 2.1) Determine the discrete density gradient and traffic isochrone center points of the target road network. The discrete density gradient is the gradient sampled from N discrete points at different proportions. It is assumed that the traffic isochrones generated from M discrete points reflect the actual isochronous state; therefore, the set of M sampled points constitutes the basic sampling dataset. The discrete density gradient proportions are selected as 0.

1. 2 M discrete points, 0.2 2 M discrete points, ..., 0.9 2 M discrete points; 2.2) Generate high-density discrete points. Use online map platforms and measured data to obtain discrete point travel time network data at a certain time as the basic sampling dataset for the study, which represents the actual observation value. Based on the idea of ​​spatial sampling, sample from the basic sampling dataset according to the gradient at a specific ratio to form sampling datasets with different discrete densities. Randomly generate discrete points with different discrete densities within the research range of the target road network. 2.3) Spatial Autocorrelation Analysis of Datasets with Different Discrete Densities: Since isochronous eigenvalues ​​are spatially continuous, they should exhibit significant spatial autocorrelation. To ensure that the generated traffic isochrones reflect the spatial distribution characteristics of isochronous eigenvalues ​​at a given moment, the discrete density sampling dataset must pass a significance test for spatial autocorrelation. The global Moran's index (MI) is used to evaluate the spatial autocorrelation of different discrete density sampling datasets. For hypothesis testing analysis of spatial autocorrelation, firstly, the Z-test score and p-value are used to determine whether the hypothesis is significant. Then, the MI value is used to determine whether the data exhibits spatial autocorrelation. The MI value is distributed in [-1, 1]. When the value is greater than 0, it indicates a positive correlation; when the value is less than 0, it indicates a negative correlation; and when the value is equal to 0, it indicates no correlation. 2.4) Estimation Error Analysis: Ordinary kriging interpolation is used to generate isochronous data. The error between the estimated values ​​and actual observed values ​​at the same observation point is calculated for isochronous data with different discrete densities. This simplifies the calculation of the estimation error of traffic isochronous lines to the calculation of the estimation error of interpolated isochronous data. The high discrete density base sampling dataset from step 2.2) is used as the benchmark value for the estimation error. The root mean square error (RMSE) and mean absolute error rate (MAER) are used to evaluate the magnitude of the estimation error, and the coefficient of variation (CV) is used to evaluate the degree of dispersion of the estimation error. In the formula: —Estimated isochronous feature values ​​of the i-th observation point in the isochronous domain generated by different discrete density sampling datasets; —The actual observed isochronous characteristic value of the i-th observation point; —The average of the estimated isochronous eigenvalues ​​for observation points with different discrete densities; n — the number of observation points; k—the sampling proportion of the sampled dataset in the base sampled dataset. When k equals 0.1, it means that 0.1 points are drawn from the M discrete points in the base sampled dataset. 2 M discrete points; 2.5) By observing the changes in travel time data MI values ​​and the relationship between discrete density and error index and coefficient of variation of different discrete gradients of the road network, a reasonable discrete density range is selected, which meets the following requirements: ① The isochronous characteristic value Z test is significant within this density range; ② Under the condition of being less than or equal to this density range, as the discrete density increases, the MI value increases significantly, and the estimated value error and its coefficient of variation decrease significantly.

4. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 3, is characterized in that: In step 3), a paired-samples t-test is performed on the estimation results of the sampled dataset. Since the spatial information carried by discrete points on roads in the road network is richer than that carried by discrete points on land parcels, the discrete rule is set as the discrete density of land parcels being less than the average discrete density of the road network. Therefore, the study only performs hypothesis testing on the density of uniformly distributed discrete points with discrete densities below the average discrete density of the road network. The null hypothesis is that there is a significant difference between the estimated and observed data. If the p-value is less than 0.1, the null hypothesis is accepted, indicating that there is a significant difference between the two data sets, meaning that the estimated value cannot correctly reflect the observed value. If the p-value is greater than 0.1, the null hypothesis is rejected, indicating that the estimated value is consistent with the observed value. The smaller the average standard error, the larger the p-value, indicating that the estimation effect is better. As the discrete density of land parcels increases, the average standard error of the paired samples gradually decreases, and the significance p-value gradually increases. At a certain density, both the significance p-value and the average standard error tend to stabilize, and this density is selected as the discrete density of the land parcels.

5. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 4, is characterized in that: In step 4), the total number of target discrete points N0 is determined, and the calculation formula is as follows: In the formula, A network For the target road network area, Let N0 be the average discrete density of the target road network, where N0 is an integer.

6. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 5, is characterized in that: In step 5), the key nodes and key road segments are identified, including the following steps: 5.1) Taking into account the topological characteristics and traffic operation characteristics of the road network, and referring to the existing research results on the evaluation indicators of key nodes of the road network, a key node identification indicator system is established from the two aspects of road network indicators and traffic indicators. 5.2) The method for determining the weights of key nodes and key road sections based on the entropy weight method is as follows: Step 1: Establish an initial matrix X, which contains n′ nodes and k′ evaluation indicators, where x fg This represents the data for the g-th evaluation index of the f-th node; X=(x fg ) n′×k′ ,f=1,2,…,n',g=1,2,…,k′ Step 2: Normalize the initial matrix X to obtain the index normalized matrix B, x min and x max These represent the minimum and maximum values ​​under the same evaluation metric. B=(b fg ) n′×k′ In the formula, b fg This represents the element of the f-th node and the g-th evaluation in the normalized matrix B; Step 3: Calculate the entropy S of the g-th evaluation index. g ; In the formula, P fg The weight of the index value of the f-th node under the g-th evaluation index; Step 4: Calculate the entropy weight ω of the g-th evaluation index. g ; 5.3) Ranking of Nodes Based on Grey Relational Analysis: Grey relational degree is the degree of similarity or dissimilarity in the development trends of factors in a system. The higher the grey relational degree, the higher the homogeneity of the change trends between the two factors, that is, the higher the degree of correlation between them. The maximum value of each evaluation index is taken to construct a reference sequence. The higher the grey relational degree between the index sequence of other nodes and the reference sequence, the higher the importance of the node. The average relational degree of the nodes connected to the road segment is calculated as the basis for ranking the importance of the road segment. The steps are as follows: Step 1: Dimensionless processing of raw indicator data: The result of the mean method can best reflect the differences in the degree of variation of the raw data of each indicator. Therefore, the mean method is used to process the raw data into dimensionless form. In the formula, This refers to the data after dimensionless processing of the data for the g-th evaluation index at the f-th node; Step 2: Construct a reference sequence Y for evaluating node importance. * ; In the formula, Let be the maximum value of the k′-th evaluation index at the f-th node after dimensionless processing, denoted as Step 3: Calculate the correlation coefficient r of each evaluation indicator. f (g); In the formula, A is the reference sequence Y evaluated by the importance of nodes. * and dimensionless data The constructed matrix This represents the data after dimensionless processing of the k′-th evaluation index data of the n′-th node, r. f (g) represents the correlation coefficient of the g-th evaluation index at the f-th intersection, Δ min and Δ max denoted as the minimum and maximum values ​​in matrix A, and ρ is the resolution coefficient; Step 4: Calculate the correlation degree R(f) of each node. The correlation degree is the importance of the node. The higher the correlation degree, the more important the node is. Step 5: Determine the correlation degree of all road segments based on the adjacency matrix; In the formula, R(ab) represents the road segment e connecting nodes a and b. ab The degree of correlation; 5.4) Determine the number of key nodes and key road segments based on a specific ratio, and rank them according to their correlation. For a road network with n′ nodes and m road segments, determine the ratio p' of key nodes and the ratio p” of key road segments, then the number of key nodes n key =round(n'×p',0), where m is the number of critical road segments. key =round(m×p”,0); Sort the node importance R(f) in ascending order to form a node importance sequence R. node ={R(f) min ,…,R(f) max }, the first n of the sequence key The nodes corresponding to each degree of correlation are the key nodes of the road network when the ratio is p', forming the key node set V. key ∈V, where V is the set of all nodes in the road network. The correlation degree R(ab) of road segments is sorted in ascending order to form the road segment importance sequence R. edge ={R(ab)} min ,...,R(ab) max }, the first m of the sequence key The road segments corresponding to each degree of correlation are the key road segments of the road network when the proportion is p”, and the set E of key road segments is formed by them. key ∈E, where E is the set of all road segments in the road network.

7. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 6, is characterized in that: In step 6), the actual total number of discrete points is calculated, including the following steps: a. Determine the actual number of discrete points on the surface: The land parcel is uniformly divided into a series of grids with a specific density. The geometric center of the grid generates discrete points on the surface of the land parcel. However, some discrete points on the surface also contain road line discrete points within their respective grids. These are duplicate discrete points on the surface and will be deleted when calculating isochronous eigenvalues. First, determine the repetition rate of the discrete points on the surface. Then, calculate the number of discrete points on the actual surface using the following formula: In the formula, k is an integer representing the number of discrete points on the actual surface. min The minimum discrete density of the land surface to ensure that isochronous interpolation does not produce holes; b. Determine the number of discrete nodes: key nodes The corresponding number of discrete points is Equal to key nodes Number of connected road segments, non-critical nodes v f The number of discrete points is 1, and the total number of discrete points in the node is N. V Its value is equal to the sum of the number of discrete points of all critical nodes and non-critical nodes; c. Determine the number of discrete points for each road segment: Divide the road segments into two categories: critical road segments and non-critical road segments. First, based on the road segment correlation score from step 5.3), determine the difference coefficient μ between the two categories of road segments. r It is the ratio of the discrete density of critical road segments to the discrete density of non-critical road segments, and is equal to the ratio of the mean correlation degree of critical road segments to that of non-critical road segments. Then, the discrete density of critical road segments and non-critical road segments are calculated separately. Finally, the number of discrete points of all actual road segments is calculated based on the discrete densities of critical and non-critical road segments. Their sum is the number of discrete points in the road segment. The calculation formula is as follows: In the formula, k is an integer; key Discrete density of critical road segments; l ab For road segment e ab Length; E key A collection of key road sections; d. Calculate the actual total number of discrete points: After adjusting the discrete points according to the discretization rules, the actual total number of discrete points N′ is the sum of the adjusted surface discrete points, node discrete points, and road segment discrete points; 8. The method for non-uniform discretization of road networks for generating isochronous traffic lines in a single center, as described in claim 7, is characterized in that: In step 7), it is determined whether the difference between the actual total number of discrete points and the target total number of discrete points is within the error range. If it is not within the error range, the process returns to step c to readjust the discrete density k of the critical road segment. key If within the error range, output the discrete density k of the critical road segment. key Given the discrete density k″ of non-critical road segments, output the non-uniform discrete scheme of the road network.

Images