A road network spatial pattern recognition method fusing road density and direction entropy

Abstract

The application discloses a road network spatial pattern recognition method fusing road density and direction entropy, belongs to the technical field of transportation planning and geographic spatial analysis, and solves the problems that existing methods are limited to a single density index, ignore structural complexity, and cannot quantize the nonlinear saturation relationship between density and direction entropy. The method comprises the following steps: calculating and analyzing the road density and direction entropy of a unit, adopting a Michaelis-Menten type saturation curve model to fit the density-entropy relationship and calculate a critical road density threshold value, preliminarily distinguishing the road network characteristics based on threshold value comparison, extracting principal components by redundant analysis dimension reduction with the road density and direction entropy as variables, and adopting K-means clustering to recognize the road network spatial pattern. The application can effectively reveal the spatial structure characteristics and regional differentiation law of the road network, and provides quantitative technical support for differentiated layout and structure optimization of the road network.

Description

Technical Field

[0001] This invention belongs to the field of transportation planning and geospatial analysis technology, and in particular relates to a method for road network spatial pattern recognition that integrates road density and directional entropy. Background Technology

[0002] Road networks are core infrastructure supporting economic and social operations and land development. Their spatial distribution and structural characteristics directly determine regional traffic efficiency and the smooth flow of resources. With the continuous construction of transportation infrastructure, the national road network has expanded rapidly. However, significant structural imbalances and heterogeneous differences still exist at the spatial level: the eastern coastal and plain areas have dense road network coverage and diverse directional layouts, with some areas exhibiting problems such as over-coverage, structural redundancy, and diminishing marginal benefits; while the mountainous and plateau regions of central and western China generally suffer from insufficient road network coverage, a simple structure, and a convergent directional layout, resulting in weak traffic accessibility and difficulty in meeting regional development needs.

[0003] Existing studies on road network spatial pattern assessment often focus on single-scale indicators such as road density, mileage, and connectivity, emphasizing the socio-economic and ecological environmental effects of road network construction. They lack a systematic analytical framework that couples coverage scale with structural characteristics, making it difficult to comprehensively depict the inherent differences in road network spatial patterns. In practical applications, relying solely on road density indicators cannot distinguish between two areas with similar densities but drastically different internal structures. For example, a city core area with a uniformly distributed, directionally balanced road network differs fundamentally from a mountainous county with a road network extending along a single river valley and highly concentrated in direction; however, a single density indicator cannot reveal this difference.

[0004] More critically, existing research has not addressed the nonlinear saturation correlation between road network density and structural complexity. Observations reveal that when road network density is low, new road construction significantly increases directional diversity and structural complexity; however, as density continues to increase, the contribution of new roads to directional entropy gradually weakens, eventually reaching a saturation limit. Current methods lack data-driven quantitative identification tools for this critical characteristic, making it difficult to accurately determine the spatial type, saturation state, and structural optimization potential of road networks in different regions. This leads to difficulties in planning decisions regarding whether a region should continue expanding its road network or shift towards optimizing its topology, hindering the scientific identification and differentiated management of large-scale, cross-regional road network spatial patterns. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a road network spatial pattern recognition method that integrates road density and directional entropy, thereby resolving the issues present in the prior art.

[0006] Firstly, to achieve the above objectives, the present invention provides a road network spatial pattern recognition method that integrates road density and directional entropy, comprising the following steps: The study area is divided into grids, the road network is divided into road segments according to intersections and endpoints, and the azimuth angle of each road segment is calculated. The azimuth angle is divided into several intervals, the road length ratio of each interval in each grid is statistically analyzed, the grid direction entropy is calculated according to the information entropy formula, and the average grid direction entropy is taken as the direction entropy of the analysis unit. Calculate the road density for each analysis unit; By using the road density and directional entropy of several analysis units, a saturation curve model is fitted to obtain the saturation curve of directional entropy as a function of road density, the theoretical maximum directional entropy is determined, and the critical road density corresponding to the maximum directional entropy at a preset ratio is calculated. By comparing the road density of each analysis unit with the critical road density, preliminary discrimination results of road network spatial characteristics are obtained. Using road density and directional entropy of each analysis unit as variables, principal components are extracted through redundancy analysis, and K-means clustering is performed on the principal component scores. Based on the clustering results, the road network spatial pattern is divided.

[0007] Optionally, the process of calculating the directional entropy of each analysis unit includes: dividing the study area into regular grids with a grid scale of 5 to 15 kilometers, and using a smaller grid scale when the study area is at the city scale and a larger grid scale when the study area is at the regional scale.

[0008] Optionally, the process of calculating the directional entropy of each analysis unit also includes: dividing the road network into discrete road segments according to intersections and endpoints, calculating the directional angle of each road segment, with the directional angle ranging from 0° to 360°; dividing the directional angle range from 0° to 360° into 24 to 72 directional statistical intervals; calculating the total road length within each directional statistical interval and the proportion of the road length in that interval to the total road length of the grid; calculating the directional entropy of each grid according to the Shannon entropy formula, and taking the average of the directional entropies of all grids within the analysis unit as the directional entropy of that unit.

[0009] Optionally, the process of calculating the road density of each analysis unit includes: obtaining the total road length and area of ​​the analysis unit, and obtaining the initial road density by dividing the total road length by the area; performing dimensionless processing on the initial road density using extreme value normalization or Z-score normalization methods to obtain the processed road density.

[0010] Optionally, the process of fitting the saturation curve model includes: using a Michaelis-Menten type saturation curve model to perform nonlinear fitting on the road density and directional entropy sample pairs of several analysis units to obtain the theoretical maximum directional entropy and the half-saturation density constant; calculating the road density corresponding to reaching 90% of the theoretical maximum directional entropy as the critical road density.

[0011] Optionally, the process of comparing the road density of each analysis unit with the critical road density to obtain the preliminary judgment result of the road network spatial characteristics includes: when the road density of the analysis unit is less than the critical road density, it is determined to be of insufficient scale type; when the road density of the analysis unit and the critical road density meet the close condition, it is determined to be of scale and structure synergistic improvement type; when the road density of the analysis unit is greater than the critical road density, it is determined to be of scale saturation type.

[0012] Optionally, the process of performing K-means clustering on the principal component scores includes: evaluating different numbers of clusters using the silhouette coefficient, variance ratio criterion, or elbow rule, determining the optimal number of clusters, and dividing the road network into four spatial patterns.

[0013] Optionally, the road network can be divided into four spatial patterns: low-density low-entropy, medium-density medium-entropy, high-density high-entropy, and high-density high-entropy; and spatial pattern labels, partition maps, and corresponding optimization strategies can be generated for each analysis unit.

[0014] Secondly, the present invention also provides a computer terminal device, comprising: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the road network spatial pattern recognition method that integrates road density and directional entropy in the first aspect described above.

[0015] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the road network spatial pattern recognition method that integrates road density and directional entropy in the first aspect described above.

[0016] Compared with the prior art, the present invention has the following advantages and technical effects: This invention provides a road network spatial pattern recognition method that integrates road density and directional entropy. By simultaneously characterizing both road density and directional entropy, it overcomes the limitations of existing methods that rely solely on a single scale indicator to evaluate road networks, and can simultaneously reflect the coverage level and structural complexity of the road network. By establishing a nonlinear saturation relationship model between density and directional entropy, this invention quantifies the critical characteristics when the road network structure tends to stabilize, providing a quantitative basis for determining whether the road network should shift from scale expansion to structural optimization. Based on this, combined with dimensionality reduction and clustering methods, it achieves accurate identification and scientific zoning of road network spatial patterns in different regions. This method can effectively reveal the spatial structural characteristics and regional differentiation patterns of road networks, providing quantitative technical support for differentiated layout and structural optimization of road networks. Attached Figure Description

[0017] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating a road network spatial pattern recognition method that integrates road density and directional entropy according to an embodiment of the present invention. Figure 2 This is a road density result diagram according to an embodiment of the present invention; Figure 3 This is a diagram showing the road direction entropy results of an embodiment of the present invention; Figure 4 This is a graph showing the fitting curve results of road density and directional entropy in an embodiment of the present invention; Figure 5 This is a diagram showing the spatial pattern partitioning results of the road network according to an embodiment of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0020] Example 1 like Figure 1 As shown, this embodiment provides a road network spatial pattern recognition method that integrates road density and directional entropy, including: The study area is divided into grids, the road network is divided into road segments according to intersections and endpoints, and the azimuth angle of each road segment is calculated. The azimuth angle is divided into several intervals, the road length ratio of each interval in each grid is statistically analyzed, the grid direction entropy is calculated according to the information entropy formula, and the average grid direction entropy is taken as the direction entropy of the analysis unit. Calculate the road density for each analysis unit; By using the road density and directional entropy of several analysis units, a saturation curve model is fitted to obtain the saturation curve of directional entropy as a function of road density, the theoretical maximum directional entropy is determined, and the critical road density corresponding to the maximum directional entropy at a preset ratio is calculated. By comparing the road density of each analysis unit with the critical road density, preliminary discrimination results of road network spatial characteristics are obtained. Using road density and directional entropy of each analysis unit as variables, principal components are extracted through redundancy analysis, and K-means clustering is performed on the principal component scores. Based on the clustering results, the road network spatial pattern is divided.

[0021] Furthermore, the process of calculating the directional entropy of each analysis unit includes: dividing the study area into regular grids with a grid scale of 5 to 15 kilometers, and using a smaller grid scale when the study area is at the city scale and a larger grid scale when the study area is at the regional scale.

[0022] Furthermore, the process of calculating the directional entropy of each analysis unit also includes: dividing the road network into discrete road segments according to intersections and endpoints, calculating the directional angle of each road segment, with the directional angle ranging from 0° to 360°; dividing the directional angle range of 0° to 360° into 24 to 72 directional statistical intervals; calculating the total road length within each directional statistical interval and calculating the proportion of the road length in that interval to the total road length of the grid; calculating the directional entropy of each grid according to the Shannon entropy formula, and taking the average of the directional entropies of all grids within the analysis unit as the directional entropy of that unit.

[0023] Specifically, the implementation process of this embodiment includes: In the grid division step, the grid scale is adaptively selected based on the scale of the study area, preferably a regular grid of 5-15km. When the study area is at the city scale, a smaller grid scale is used, and when the study area is at the regional or metropolitan area scale, a larger grid scale is used to improve the stability of directional entropy calculation and the accuracy of road network structure feature identification.

[0024] The road network is divided into discrete road segments according to intersections and endpoints, and the direction angle of each road segment is calculated, ranging from 0° to 360°. The 0°–360° direction angle range is further divided into multiple direction statistical intervals, preferably 24–72 intervals, with a typical division method being 36 equally divided 10° direction intervals. The total road length within each direction statistical interval is calculated, and the proportion P(θ) of the road length within that interval to the total road length of the corresponding grid is calculated. The direction entropy of each grid is calculated, and the average pixel entropy of all grids within the analysis unit is taken as the direction entropy S of that unit, used to characterize the direction diversity, structural complexity, and connectivity of the road network. A higher entropy value indicates a more complex and sustainable road network structure. The Shannon entropy formula is used to calculate the pixel entropy of each grid. ; in, Si Represents a grid i Road entropy, n P(represents the total number of directions counted) Oθ ) indicates that it is located θ Percentage of road length within a directional range.

[0025] Furthermore, the process of calculating the road density of each analysis unit includes: obtaining the total road length and area of ​​the analysis unit, and obtaining the initial road density by dividing the total road length by the area; and performing dimensionless processing on the initial road density using extreme value normalization or Z-score normalization methods to obtain the processed road density.

[0026] Specifically, the implementation process of this embodiment includes: The calculation of the area and total road length of each analysis unit includes: First, determining analysis units with clear spatial boundaries, such as county-level administrative regions, prefecture-level administrative regions, or regular grids, based on the research objectives, and assigning a unique identifier to each analysis unit; Second, uniformly converting the analysis unit data and road network data to a plane projection coordinate system suitable for the research area, calculating the polygon area of ​​each analysis unit in this coordinate system, and uniformly converting it to square kilometers; Subsequently, performing duplicate element checks, invalid geometry repair, and topological relationship processing on the road network data, and using the analysis unit boundaries to spatially clip or overlay roads. Road segments entirely within a single analysis unit are all included in that unit, while roads spanning multiple analysis units are divided according to the analysis unit boundaries, and the road segments within different analysis units are assigned to their respective units; Finally, calculating the planar geometric length of each road segment, and grouping and accumulating them according to the unique identifier of the analysis unit to obtain the length of the first segment. i The total road length of each analysis unit is obtained. This yields the area and total road length of each analysis unit, serving as the basis for subsequent road density calculations.

[0027] Calculate the road density of each analysis unit to characterize the road network size; ; in, RD i Road density (km / km) of analysis unit i 2 ), RL i This represents the total road length (km) of analysis unit i. A i Represents the area of ​​analysis unit i; The road density of each analysis unit is dimensionless by using extreme value normalization or Z-score standardization to eliminate the influence of differences in area scale between different regions on the results.

[0028] Furthermore, the process of fitting the saturation curve model includes: using the Michaelis-Menten type saturation curve model to perform nonlinear fitting on the road density and directional entropy sample pairs of several analysis units to obtain the theoretical maximum directional entropy and the half-saturation density constant; calculating the road density corresponding to reaching 90% of the theoretical maximum directional entropy, as the critical road density.

[0029] Specifically, the implementation process of this embodiment includes: Collect (RD, S) sample pairs from multiple analysis units to form a road density-directional entropy sample dataset; The Michaelis-Menten saturation curve model was used to perform nonlinear fitting on the (RD,S) sample to obtain the theoretical maximum entropy S_max and the half-saturation density constant. The critical road density threshold RD_threshold corresponding to 90% of the theoretical maximum entropy S_max was calculated as the key turning point for identifying the road network from scale expansion to structural optimization.

[0030] The Michaelis-Menten type saturation curve model was used to perform nonlinear fitting on the sample dataset to obtain the theoretical maximum entropy S_max and the half-saturation density constant; ; in, S m Represents the theoretical maximum entropy, k It is the road density corresponding to the maximum entropy, also known as the half-saturation density constant; Calculate the critical road density threshold RD_threshold corresponding to the maximum entropy S_max that reaches the preset ratio, and use it as the key discrimination threshold for the diminishing marginal benefits of road network expansion and the optimization of turning structure. ; inP This is a preset ratio.

[0031] Furthermore, the process of comparing the road density of each analysis unit with the critical road density to obtain the preliminary judgment result of the road network spatial characteristics includes: when the road density of the analysis unit is less than the critical road density, it is determined to be of insufficient scale type; when the road density of the analysis unit and the critical road density meet the close condition, it is determined to be of scale and structure synergistic improvement type; when the road density of the analysis unit is greater than the critical road density, it is determined to be of scale saturation type.

[0032] Specifically, the implementation process of this embodiment includes: For each analysis unit, compare its current road density (RD) with the calculated critical road density threshold (RD_threshold).

[0033] Preliminary identification of road network spatial characteristics based on relative relationships: If RD is much lower than RD_threshold, it is judged as insufficient in scale (significant potential for increasing structural complexity). If RD is close to RD_threshold, it is determined to be a type of scale and structure synergistic improvement. If RD exceeds RD_threshold, it is determined to be a saturated scale (the entropy gain is marginally decreasing, and structural optimization should be carried out first).

[0034] Furthermore, the process of K-means clustering of principal component scores includes: evaluating different numbers of clusters using silhouette coefficients, variance ratio criteria, or elbow rule, determining the optimal number of clusters, and dividing the road network into four spatial patterns.

[0035] Furthermore, the road network is divided into four spatial patterns: low-density low-entropy, medium-density medium-entropy, high-density high-entropy, and high-density high-entropy; and spatial pattern labels, partition maps, and corresponding optimization strategies are generated for each analysis unit.

[0036] Specifically, the implementation process of this embodiment includes: Road density (RD) and directional entropy (S) were used as comprehensive feature variables for data standardization.

[0037] Redundancy analysis (RDA) was used to rank RD and S, and the two variables were transformed into linear disjoint principal components (the first 2 to 3 principal components were extracted to ensure that the explained variance was >80%).

[0038] Redundancy analysis is a constrained ranking method that combines multiple linear regression with principal component analysis. In this embodiment, its specific application is as follows: First, a road network characteristic response matrix is ​​constructed using road density (RD) and directional entropy (S), and a constrained variable matrix is ​​constructed using relevant attributes that can explain the spatial differences in the road network among different analysis units. Second, multiple linear regression models of road density and directional entropy relative to the constrained variables are established to obtain the fitted values ​​of RD and S, which are then used to form a constrained road network characteristic matrix. Subsequently, the covariance matrix and eigenvalue decomposition are performed on the constrained road network characteristic matrix, converting RD and S into several mutually orthogonal and linearly uncorrelated RDA ranking axes. Finally, the ranking order is determined based on the magnitude of the eigenvalues ​​of each ranking axis, and the main ranking axes are selected according to the cumulative explained variance. The scores of each analysis unit on the main ranking axes are used as input for subsequent K-means clustering.

[0039] Using principal component scores as input data, the K-means clustering algorithm is employed to cluster the analysis units. The specific process is as follows: First, the principal component scores of each analysis unit are used to form a clustering input matrix, where each row represents an analysis unit and each column represents a principal component score. Second, a preset number of clusters K is used, and K initial cluster centers are selected using the K-means method. The Euclidean distance between each analysis unit and different cluster centers is calculated, and the unit is assigned to the nearest cluster. Subsequently, the average principal component scores of all analysis units in each cluster are calculated, the cluster centers are updated, and the process of "distance calculation, category division, and cluster center update" is repeated until the category affiliation of the analysis unit no longer changes, or the change in cluster centers is less than a preset threshold. To reduce the impact of random selection of initial cluster centers, the clustering process is repeated multiple times, and the result with the smallest sum of squared errors within each cluster is selected as the final clustering result. Finally, each analysis unit is assigned a corresponding category label, and the road network spatial pattern type is determined based on the road density RD and directional entropy S characteristics corresponding to each category, generating the road network spatial pattern partitioning result.

[0040] The optimal number of clusters was determined using the silhouette coefficient, variance ratio criterion, and elbow rule. The specific process is as follows: First, a range of candidate cluster numbers K was defined, and K-means clustering was performed for different K values. Then, the average silhouette coefficient, Calinski-Harabasz exponent, and sum of squared errors (SSE) of each cluster were calculated. Larger silhouette coefficients and Calinski-Harabasz exponents indicate more significant intra-cluster similarity and inter-cluster differences, resulting in better clustering performance. Simultaneously, a curve showing the change in K value versus SSE was plotted, with the point where the SSE decreases from a significant point to a gradual flattening point designated as the elbow point. By comprehensively comparing the above results, the K value with a larger silhouette coefficient and Calinski-Harabasz exponent, and close to the elbow point of the SSE curve, was selected as the optimal number of clusters. When evaluation results are inconsistent, the number of clusters supported by most indicators and showing clear differences in road network characteristics among the categories was prioritized.

[0041] The clustering results are divided into four typical road network spatial patterns: low density and low entropy, medium density and medium entropy, high density and high entropy, and high density and high entropy. Spatial pattern labels, spatial partition maps and differentiated optimization strategies are generated for each analysis unit, including prioritizing the improvement of coverage density and connectivity in areas with insufficient scale, and focusing on optimizing the road network topology and directional balance in areas with saturated scale.

[0042] Example 2 In this embodiment, a computer terminal device is provided, including: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described method for road network spatial pattern recognition that integrates road density and directional entropy.

[0043] In this embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described method for road network spatial pattern recognition that integrates road density and directional entropy.

[0044] An application example of this invention is as follows: Several county-level administrative regions in a province were selected as experimental subjects, and road network data from a specific year was used for research and analysis. The road network data includes attributes such as road grade, node distribution, connectivity, and road length, effectively reflecting the spatial distribution and structural characteristics of the county-level road network in the province. Using a 5km×5km grid as the basic computational unit, the OSM road vector data was first preprocessed (topology checking, directional segmentation, and 30m rasterization). Then, the road density (RD) and directional entropy (S) for each county-level administrative region were calculated. Based on multi-region (RD, S) sample pairs, a Michaelis-Menten type saturation curve model was used for fitting to obtain the theoretical maximum entropy S_max and the half-saturation density constant k, and the critical road density threshold RD_threshold at 90% maximum entropy was calculated. Subsequently, a preliminary diagnosis of the road network spatial characteristics was performed based on the relative relationship between the current RD and RD_threshold for each region. Finally, using RD and S as comprehensive feature variables, principal components were extracted through redundancy analysis (RDA) for dimensionality reduction, and K-means clustering was used to determine the optimal number of clusters, subdividing the road network into different spatial patterns.

[0045] Figure 2 This is a spatial distribution map of road density in county-level administrative regions of a province. Using several county-level administrative regions as analytical units, the map shows the spatial differences in the total road length per unit area within each region. Different levels in the map represent different road density levels; higher road density indicates a higher degree of road network coverage in the area.

[0046] Figure 3 This is a spatial distribution map of road directional entropy in a province's county-level administrative regions. The map illustrates the spatial differences in the diversity and balance of road directions across these regions. Higher directional entropy values ​​indicate a richer and more balanced distribution of roads in different directions; lower values ​​indicate that roads are mainly concentrated in a few directions, resulting in a relatively simple directional structure.

[0047] Figure 4 This is a nonlinear fitting graph showing the relationship between road density and road directional entropy. The graph uses road density (RD) as the x-axis and road directional entropy (S) as the y-axis. The scattered points in the graph represent different county-level administrative regions, and the fitted curve represents the overall trend of directional entropy as road density increases. Based on the fitting results, the critical road density threshold corresponding to when directional entropy tends to stabilize can be further identified.

[0048] Figure 5 This is a spatial pattern zoning map of the road network in a province's county-level administrative regions. Based on the comprehensive characteristics of road density and directional entropy in each county-level administrative region, the map uses ranking analysis and K-means clustering to classify different road network spatial patterns. Different categories in the map represent different combinations of road scale and directional structure, reflecting the distribution location and regional differences of various road network spatial patterns.

[0049] This embodiment can intuitively demonstrate the stage-specific characteristics and saturation differences of the road network in different zones of a province's county-level administrative region, providing quantitative support for its road network structure optimization and differentiated planning.

[0050] This invention provides a road network spatial pattern recognition method that integrates road density and directional entropy. By simultaneously characterizing both road density and directional entropy, it overcomes the limitations of existing methods that rely solely on a single scale indicator to evaluate road networks, and can simultaneously reflect the coverage level and structural complexity of the road network. By establishing a nonlinear saturation relationship model between density and directional entropy, this invention quantifies the critical characteristics when the road network structure tends to stabilize, providing a quantitative basis for determining whether the road network should shift from scale expansion to structural optimization. Based on this, combined with dimensionality reduction and clustering methods, it achieves accurate identification and scientific zoning of road network spatial patterns in different regions. This method can effectively reveal the spatial structural characteristics and regional differentiation patterns of road networks, providing quantitative technical support for differentiated layout and structural optimization of road networks.

[0051] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A road network spatial pattern recognition method that integrates road density and directional entropy, characterized in that, include: The study area is divided into grids, the road network is divided into road segments according to intersections and endpoints, and the azimuth angle of each road segment is calculated. The azimuth angle is divided into several intervals, the road length ratio of each interval in each grid is statistically analyzed, the grid direction entropy is calculated according to the information entropy formula, and the average grid direction entropy is taken as the direction entropy of the analysis unit. Calculate the road density for each analysis unit; By using the road density and directional entropy of several analysis units, a saturation curve model is fitted to obtain the saturation curve of directional entropy as a function of road density, the theoretical maximum directional entropy is determined, and the critical road density corresponding to the maximum directional entropy at a preset ratio is calculated. By comparing the road density of each analysis unit with the critical road density, preliminary discrimination results of road network spatial characteristics are obtained. Using road density and directional entropy of each analysis unit as variables, principal components are extracted through redundancy analysis, and K-means clustering is performed on the principal component scores to divide the road network spatial pattern based on the clustering results.

2. The method according to claim 1, characterized in that, The process of calculating the directional entropy of each analysis unit includes: dividing the study area into regular grids with a grid scale of 5 to 15 kilometers, and using a smaller grid scale when the study area is at the city scale and a larger grid scale when the study area is at the regional scale.

3. The method according to claim 2, characterized in that, The process of calculating the directional entropy of each analysis unit also includes: dividing the road network into discrete road segments according to intersections and endpoints, calculating the directional angle of each road segment, with the directional angle ranging from 0° to 360°; dividing the directional angle range of 0° to 360° into 24 to 72 directional statistical intervals; calculating the total road length within each directional statistical interval and the proportion of the road length in that interval to the total road length of the grid; calculating the directional entropy of each grid according to the Shannon entropy formula, and taking the average of the directional entropies of all grids within the analysis unit as the directional entropy of that unit.

4. The method according to claim 1, characterized in that, The process of calculating the road density of each analysis unit includes: obtaining the total road length and area of ​​the analysis unit, and obtaining the initial road density by dividing the total road length by the area; and performing dimensionless processing on the initial road density using extreme value normalization or Z-score normalization methods to obtain the processed road density.

5. The method according to claim 1, characterized in that, The process of fitting the saturation curve model includes: using the Michaelis-Menten type saturation curve model to perform nonlinear fitting on the road density and directional entropy sample pairs of several analysis units to obtain the theoretical maximum directional entropy and the half-saturation density constant; calculating the road density corresponding to reaching 90% of the theoretical maximum directional entropy, which is taken as the critical road density.

6. The method according to claim 1, characterized in that, The process of comparing the road density of each analysis unit with the critical road density to obtain the preliminary judgment result of the road network spatial characteristics includes: when the road density of the analysis unit is less than the critical road density, it is determined to be of insufficient scale type; when the road density of the analysis unit and the critical road density meet the close condition, it is determined to be of scale and structure synergistic improvement type; when the road density of the analysis unit is greater than the critical road density, it is determined to be of scale saturation type.

7. The method according to claim 1, characterized in that, The process of K-means clustering of principal component scores includes: evaluating different numbers of clusters using silhouette coefficient, variance ratio criterion or elbow rule, determining the optimal number of clusters, and dividing the road network into four spatial patterns.

8. The method according to claim 7, characterized in that, The road network is divided into four spatial patterns: low-density low-entropy, medium-density medium-entropy, high-density high-entropy, and high-density high-entropy; and spatial pattern labels, partition maps, and corresponding optimization strategies are generated for each analysis unit.

9. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-8.

Images