GIS-based port economy spatial distribution analysis method and system

Abstract

The application discloses a GIS-based port economic space distribution analysis method and system, and relates to the technical field of data analysis. The method comprises the following steps: obtaining multi-source heterogeneous data interactively, forming a standardized data set after coordinate unification and scale normalization; performing data set gridding processing, constructing a multi-layer space system, and generating a comprehensive distribution grid field through weighting, superposition and fusion; constructing a spatial adjacency matrix based on the grid field, calculating the aggregation correlation strength, and generating a self-correlation feature map; identifying hot spots of the feature map and aggregating partitions to obtain partition results; and calculating the aggregation barycenter migration of the multi-time-phase partition results to construct an evolution path sequence. The technical problems that the port economic space distribution characteristics are difficult to accurately identify and the aggregation evolution trend is difficult to quantitatively analyze due to the inconsistency of the coordinate datum, spatial scale and time scale of the port economic related multi-source heterogeneous data are solved, and the technical effects of improving the port economic space distribution identification precision and the accuracy of the spatial aggregation evolution trend analysis are achieved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a method and system for analyzing the spatial distribution of port economic activity based on GIS. Background Technology

[0002] As a hub for both sea and land logistics, the spatial distribution characteristics of ports directly determine the layout efficiency of port-related industries and the level of cross-regional economic synergy. The industry as a whole is gradually shifting from the past stage of extensive scale expansion to a new stage of refined spatial governance and port-industry-city coordinated development. The demand for quantitative analysis of the port economy's development trend has also extended from the traditional macro-statistical dimension to a more precise micro-spatial dimension. For a long time, the analysis of port economy has mostly relied on non-spatial statistical reports, lacking a unified spatial fusion processing mechanism for multi-source heterogeneous data such as port economic added value, distribution of port-related enterprises, transportation networks, and land use. Data from different sources often suffer from problems such as inconsistent coordinate systems, misaligned spatial scales, and misaligned time slices. A large amount of industrial linkages and agglomeration evolution information hidden in the geographic space cannot be effectively mined. Traditional analysis methods are unable to accurately depict the actual development intensity of the port economy in different spatial units, nor can they dynamically track the migration path of the port economic agglomeration core. Ultimately, this leads to a significant deviation between the analysis results and the actual industrial layout, making it difficult to provide highly reliable spatial decision-making support for practical work such as port planning, port-related industry investment promotion, and transportation network optimization. Summary of the Invention

[0003] This application provides a GIS-based method and system for analyzing the spatial distribution of port economy. The key feature is that it solves the technical problem that the spatial distribution characteristics of port economy are difficult to accurately identify and the clustering and evolution trend is difficult to quantitatively analyze due to the inconsistency in coordinate benchmarks, spatial scales and temporal scales of multi-source heterogeneous data related to port economy. It achieves the technical effect of improving the accuracy of identifying the spatial distribution of port economy and the accuracy of analyzing the spatial clustering and evolution trend by standardizing the spatial representation of multi-source heterogeneous data, layer overlay and fusion, spatial autocorrelation analysis and multi-temporal evolution modeling.

[0004] The first aspect of this application provides a GIS-based method for analyzing the spatial distribution of port economic activity, the method comprising: Interactive acquisition of multi-source heterogeneous data, including port economic added value data, port-adjacent enterprise distribution data, transportation network and land use data, is performed. After coordinate unification and scale normalization, the multi-source heterogeneous data forms a standardized spatial element dataset. This standardized spatial element dataset is then rasterized to construct a multi-layer spatial representation system with unified resolution. Weight assignment and overlay fusion of each element layer are performed within this multi-layer spatial representation system to generate a comprehensive port economic spatial distribution raster field. A spatial adjacency matrix is ​​constructed based on this comprehensive port economic spatial distribution raster field. The clustering correlation strength between spatial units is calculated based on local spatial statistical indicators within the spatial adjacency matrix, generating a spatial autocorrelation distribution feature map. Hotspot identification and clustering partitioning are performed on the spatial autocorrelation distribution feature map, establishing identification partitioning results. The geometric centroid migration trajectory of the clustering center is calculated from the identification partitioning results of the multi-temporal raster field, constructing a spatial evolution path sequence.

[0005] A second aspect of this application provides a GIS-based port economic spatial distribution analysis system, the system comprising: The data standardization module interactively acquires multi-source heterogeneous data, including port economic value-added data, port-adjacent enterprise distribution data, transportation network data, and land use data. This multi-source heterogeneous data is standardized into a spatial element dataset after coordinate unification and scale normalization. The element fusion module rasterizes the standardized spatial element dataset, constructs a multi-layer spatial expression system with unified resolution, and performs weight assignment and overlay fusion of each element layer within the multi-layer spatial expression system to generate a comprehensive port economic spatial distribution raster field. The correlation analysis module constructs a spatial adjacency matrix based on the comprehensive port economic spatial distribution raster field. Within the spatial adjacency matrix, it calculates the clustering correlation strength between spatial units based on local spatial statistical indicators, generating a spatial autocorrelation distribution feature map. The clustering partitioning module identifies hotspots and partitions clusters within the spatial autocorrelation distribution feature map, establishing identification and partitioning results. The evolution analysis module calculates the geometric centroid migration trajectory of the clustering center based on the identification and partitioning results of the multi-temporal raster field, constructing a spatial evolution path sequence.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, various types of spatial data related to the port economy were collected, and these data underwent coordinate system transformation, scale conversion, and normalization to enable analysis of data from different sources and formats under the same spatial reference. Then, the processed spatial element data was converted into a raster layer with a uniform resolution. Weights were assigned and the data were overlaid and fused according to the influence of various elements on the spatial distribution of the port economy, resulting in a spatial distribution raster field reflecting the comprehensive strength of the port economy. Next, a spatial adjacency matrix was established based on the spatial proximity relationships between raster cells, and the degree of clustering correlation between different spatial cells was calculated using local spatial statistical indicators, forming an autocorrelation distribution characteristic map of the port economy. Then, based on this characteristic map, hotspot areas of port economic activity were identified, and areas with spatial continuity and clustering characteristics were partitioned to obtain the port economic spatial clustering zoning results. Finally, based on the raster analysis results at different times, the changing trajectory of the center position of each clustering area was calculated, forming a sequence of port economic spatial evolution paths, thereby achieving dynamic analysis of the expansion, transfer, and development trends of the port economy. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of the GIS-based port economic spatial distribution analysis method provided in this application embodiment.

[0009] Figure 2 This is a schematic diagram of the spatial autocorrelation distribution characteristics provided in an embodiment of this application.

[0010] Figure 3 This is a schematic diagram illustrating the adaptive adjustment verification of neighborhood window parameters provided in an embodiment of this application.

[0011] Figure 4 A schematic diagram of the structure of a GIS-based port economic spatial distribution analysis system provided in this application embodiment.

[0012] Figure labeling: Data standardization module 11, element fusion module 12, correlation analysis module 13, clustering and partitioning module 14, evolution analysis module 15. Detailed Implementation

[0013] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0014] Example 1, as Figure 1 As shown, this application provides a GIS-based method for analyzing the spatial distribution of port economic activity, wherein the method includes: Interactive acquisition of multi-source heterogeneous data, including port economic added value data, port-adjacent enterprise distribution data, transportation network and land use data, and the multi-source heterogeneous data are converted into a standardized spatial element dataset after coordinate unification and scale normalization.

[0015] In this embodiment, the GIS data import interface receives user input regarding the research area boundary, analysis year, and unified raster resolution. It then imports the port economic value added table, port-adjacent enterprise location table, transportation network vector layer, and land use vector layer according to preset data templates. The port economic value added table uses port cities as the recording unit; the port-adjacent enterprise location table uses enterprises as the recording unit and includes at least enterprise number, longitude, latitude, industry type, and annual output value fields; the transportation network vector layer uses line elements to represent road centerlines and includes a road grade field; and the land use vector layer uses polygon elements to represent land parcel boundaries and includes a land use type field. After data reception, the research area is uniformly set as the target projection coordinate system. Using this target projection coordinate system as a reference, the latitude and longitude coordinates in the port-adjacent enterprise location table are projected and transformed. The transportation network vector layer and land use vector layer are reprojected, and the port economic value added table is attribute-associated according to the city spatial boundary, thus ensuring all data falls under the same spatial coordinate reference. Subsequently, a regular raster grid covering the study area was established according to the user-defined unified spatial resolution and raster size. Using the raster cells as the unified spatial scale, the port's economic added value was allocated to corresponding raster cells, the annual output value of port-adjacent enterprises was aggregated to their respective raster cells based on enterprise location, traffic density values ​​for the transportation network were calculated based on road lengths within the raster, and land use types were converted to raster suitability values ​​according to a preset land use suitability assignment table. After completing the spatial scale conversion, range normalization was performed on the port's economic added value, aggregated enterprise output value, traffic density value, and land use suitability value in each raster cell, uniformly mapping all element values ​​to a numerical range of 0 to 1. For raster cells without corresponding element values ​​within the same analysis year, their standardized value was assigned to 0. After the above processing, a standardized spatial element dataset is formed, consisting of a standardized layer of port economic added value, a standardized layer of port-adjacent enterprise intensity, a standardized layer of transportation accessibility, and a standardized layer of land use suitability. This standardized spatial element dataset has a unified coordinate system, a unified raster scale, a unified year caliber, and a unified numerical dimension, and can serve as the data basis for subsequent multi-layer overlay and fusion, spatial autocorrelation analysis, and cluster evolution calculation.

[0016] Furthermore, the multi-source heterogeneous data, after coordinate unification and scale normalization, forms a standardized spatial element dataset, including: Spatial geographic registration and projection transformation are performed on the port economic value-added data and the port-adjacent enterprise distribution data to place the port economic value-added data, port-adjacent enterprise distribution data, traffic routes and land use data in the same geographic coordinate system; the data of each element are resampled and time sliced ​​according to a unified spatial resolution and raster size to eliminate scale differences and spatiotemporal misalignment, and a standardized spatial element dataset is established.

[0017] Preferably, the coordinate system of the land use data is first used as the target geographic coordinate system, and the spatial reference information of the traffic route data and land use data is read to determine the target projection parameters. For the distribution data of port-adjacent enterprises, the longitude and latitude fields in the enterprise location table are read, and each enterprise record is generated into a point feature. Then, a projection transformation is performed according to the target geographic coordinate system so that the enterprise locations fall into spatial positions consistent with the traffic route data and land use data. For the port economic value added data, the values ​​are read, and a pre-stored port city boundary layer is called. The port economic value added data is then attribute-linked to the corresponding city boundary according to the port city number, transforming the port economic value added data, which originally lacked a continuous spatial form, into area feature data with spatial boundaries. Subsequently, a projection transformation is performed on the surface feature data. That is, the original spatial reference parameters of the port city boundary layer are read, the original coordinate values ​​of each boundary node in the surface feature are obtained, a coordinate transformation relationship is established based on the original spatial reference parameters and the target projection parameters, and the datum transformation, projection forward calculation and coordinate reconstruction are performed on each boundary node in sequence to convert the original coordinate values ​​into planar coordinate values ​​under the target geographic coordinate system. The spatial geometry of the port city surface feature is regenerated according to the transformed boundary nodes, so that the port economic added value data, port-adjacent enterprise distribution data, traffic route data and land use data are all under the same target geographic coordinate system.

[0018] After coordinate unification, a regular raster grid is generated within the boundary of the study area according to the user-defined unified spatial resolution, and this raster grid serves as the unified spatial analysis unit. The system spatially overlays the port economic value-added surface elements with the raster grid, and allocates the port economic value-added values ​​to the corresponding raster units according to the overlap area ratio between the raster unit and the port city boundary; it places the port-adjacent enterprise point elements into the corresponding raster units, and sums the annual output values ​​of enterprises within the same raster unit to obtain the enterprise economic intensity value; it performs intersection calculations on the traffic path line elements with the raster grid, calculates the traffic path length within each raster unit, and calculates the traffic density value using the raster area as the denominator; it overlays the land use surface elements with the raster grid, determines the land use type of the raster unit according to the land use type with the largest proportion within the raster unit, and converts it into a land use suitability value according to the preset land use suitability assignment table. During time slice alignment, the system uses the analysis year as a unified time caliber, extracting the port economic value added records and port-related enterprise records corresponding to that year, and fixing the traffic route data and land use data to the versions corresponding to that analysis year. When the update time of the traffic route data and land use data is earlier than the analysis year, the data version closest to and no later than the analysis year is selected as the corresponding time slice. Then, range normalization is performed on the port economic value added, enterprise economic intensity value, traffic density value, and land use suitability value in each raster cell, converting data of different dimensions into standardized values ​​between 0 and 1. After the above resampling, time slice alignment, and normalization processes, a standardized spatial element dataset expressed in the same coordinate system, raster size, analysis year, and numerical dimensions is obtained, thereby improving the consistency and comparability of multi-source port economic data in spatial overlay analysis.

[0019] The standardized spatial element dataset is rasterized to construct a multi-layer spatial representation system with a unified resolution. In the multi-layer spatial representation system, the weights of each element layer are assigned and superimposed and merged to generate a comprehensive distribution raster field of port economic space.

[0020] In one embodiment, the processing range is defined by the boundary of the study area in the standardized spatial element dataset. A regular raster grid is established according to a preset uniform raster resolution, and each raster cell is used as the smallest computational unit for port economic spatial analysis. The system reads the standardized elements of port economic added value, port-adjacent enterprise intensity, transportation accessibility, and land use suitability, respectively, and converts each type of element into a corresponding raster layer according to the raster cell number. For the port economic added value element, the port urban surface element is overlaid with the raster grid, and the standardized value of urban port economic added value is allocated to the corresponding raster cell according to the overlap area ratio between the raster cell and the urban surface element. For the port-adjacent enterprise intensity element, the enterprise location is placed in the raster cell, and the standardized values ​​of enterprise output within the same raster cell are accumulated to form the enterprise intensity raster value. For the transportation accessibility element, the length of the traffic path within each raster cell is calculated, and its standardized result is used as the transportation accessibility raster value. For the land use suitability element, the land use type corresponding to each raster cell is read, and the land use suitability raster value is generated according to a preset suitability assignment table. This results in a multi-layered spatial representation system with the same number of rows and columns, the same raster size, the same spatial range, and the same coordinate reference. Subsequently, weights are assigned to each element layer. Initial weight coefficients are set for the port economic value-added layer, the port-adjacent enterprise strength layer, the transportation accessibility layer, and the land use suitability layer. An information entropy correction factor is calculated based on the dispersion of the raster values ​​in each layer, and the initial weight coefficients are normalized to obtain the final weight coefficients for each element layer. After determining the weights, the multiple layers are weighted and superimposed pixel-by-pixel according to the same raster unit position to obtain the comprehensive port economic strength value of that raster unit. Finally, the comprehensive strength values ​​of all raster units are normalized to the range of 0 to 1 and extreme value suppression is applied to generate a comprehensive distribution raster field of port economic space that can continuously express the differences in the strength of port economic space. This provides a unified rasterized computational basis for subsequent spatial adjacency matrix construction, spatial autocorrelation analysis, and clustering identification, improving the continuity of multi-source element fusion expression and the computability of subsequent spatial correlation analysis.

[0021] Furthermore, the generation of a comprehensive spatial distribution grid field for port economic development includes: Normalization was performed on the port economic value-added layer, port-adjacent enterprise distribution layer, transportation network layer, and land use layer, transforming the elements of each layer into rasterized numerical expressions with the same dimension. Based on the driving contribution of each layer to port economic development, weight coefficients for economic value-added, industry density, transportation accessibility, and land use suitability were initialized. The entropy method was used to calculate the dispersion of raster values ​​in each layer, and information entropy weight correction factors for each layer were established. The initialized weight coefficients were dynamically corrected using the information entropy weight correction factors, and weight combinations were established to complete the weight assignment. Based on the weight assignment, a pixel-by-pixel weighted overlay operation was performed on each element layer, and the operation results were subjected to nonlinear stretching and extreme value suppression processing to generate a comprehensive spatial distribution raster field of port economy.

[0022] Preferably, a unified raster grid is used as the calculation basis. The original feature values ​​of each raster cell in the port economic value-added layer, the port-adjacent enterprise distribution layer, the transportation network layer, and the land use layer are read respectively, and the port economic value-added is recorded as... The economic intensity value of enterprises in the port area is recorded as Traffic density value is recorded as The land use suitability value is recorded as Where i represents the i-th raster cell. Subsequently, range normalization is applied to the four types of elements. That is, using the maximum and minimum values ​​of all raster cells within a layer as the normalization benchmark, the original value of each raster cell is subtracted from the minimum value of that layer, and then divided by the difference between the maximum and minimum values ​​of that layer to obtain the corresponding standardized raster value. When the maximum and minimum values ​​of a layer are equal, the standardized value of all valid raster cells within that layer is uniformly set to 1, and the standardized value of raster cells with no valid data is set to 0. This results in standardized layers for port economic value added, industry density, transportation accessibility, and land use suitability. Then, based on the driving effect of the four types of elements on the spatial distribution of port economy, initial weights are set through expert decision-making. For example, the initial weight for port economic value added can be set to 0.35, the initial weight for industry density can be set to 0.30, the initial weight for transportation accessibility can be set to 0.20, and the initial weight for land use suitability can be set to 0.15. After initial weight settings are completed, the system uses the entropy method to calculate the information entropy weight correction factor for each layer. Specifically, for each feature layer, the proportion of the standardized value of the i-th raster cell in that layer to the sum of the standardized values ​​of all raster cells in that layer is first calculated. Then, the information entropy value of that layer is calculated based on this proportion. The smaller the information entropy value, the more obvious the differences in raster values ​​in that layer, the stronger its role in distinguishing spatial distribution, and the larger the corresponding information entropy weight correction factor. The system multiplies the information entropy weight correction factor of each layer with the corresponding initial weight and normalizes the multiplication result so that the sum of the four final weight coefficients equals 1, thus obtaining the final weight combination of the port economic value added layer, industry density layer, transportation accessibility layer, and land use suitability layer. Then, a pixel-by-pixel weighted overlay operation is performed on the four standardized layers according to the raster cell position. For the i-th raster cell, the standardized values ​​of port economic value added, industry density, transportation accessibility, and land use suitability are multiplied by their respective final weights and then summed to obtain the initial value of the comprehensive port economic intensity of that raster cell. Finally, the initial comprehensive intensity values ​​of all raster cells are subjected to nonlinear stretching, using a square root stretching method to enhance the expression of differences between low and medium value regions. The stretched results are then subjected to extremum suppression, truncating raster values ​​above the 95th quantile to their 95th quantile counterparts and renormalizing them to the 0-1 range, ultimately generating a raster field representing the comprehensive spatial distribution of port economic activity. This process preserves the combined influence of port economic added value, industrial agglomeration, transportation support, and land use suitability while reducing the interference of abnormally high values ​​on the overall spatial distribution assessment, thus improving the stability and comparability of the expression of port economic spatial intensity.

[0023] A spatial adjacency matrix is ​​constructed based on the comprehensive spatial distribution grid field of port economy. The clustering correlation strength between spatial units is calculated based on local spatial statistical indicators in the spatial adjacency matrix, and a spatial autocorrelation distribution feature map is generated.

[0024] In one embodiment, each effective grid cell in the port economic spatial distribution grid field is used as a spatial calculation unit, and the row and column numbers, center point coordinates, and port economic comprehensive intensity value of each grid cell are read. The system determines the spatial proximity relationship between grid cells one by one according to preset adjacency rules. For two grid cells that are adjacent, the basic spatial weight is calculated based on the distance between the center points of the two grid cells, and the weight is written into the corresponding row and column position in the spatial adjacency matrix, thereby forming a spatial adjacency matrix with grid cells as rows and adjacent grid cells as columns. Subsequently, the average value of the port economic comprehensive intensity value of all effective grid cells is calculated, and the deviation of the comprehensive intensity value of the target grid cell from the average value and the deviation of the comprehensive intensity value of its adjacent grid cells from the average value are calculated respectively, and a local spatial deviation vector is established based on this. Subsequently, all local spatial deviation vectors corresponding to the same target grid cell are fused element-by-element to obtain a correlation coefficient matrix that characterizes the spatial dependence strength between cells. The correlation coefficient matrix is ​​then rewritten back into the spatial grid according to the original grid row and column positions, and graded and color-mapped according to the intensity value to generate a spatial autocorrelation distribution feature map. This visually expresses the local clustering relationship and spatial dependence strength between port economic spatial cells, improving the accuracy of identifying port economic hotspots and transition areas.

[0025] Furthermore, based on the matching results, the industry added value dataset is multiplied by the industry water transport dependence coefficient to calculate the economic contribution value of ports in multiple industries. The method includes: The port economic spatial distribution grid field is divided into regular grid cells with a fixed spatial step size. A neighborhood window with a preset radius is established around each grid cell. The set of adjacent grid cells within the neighborhood window that have boundary contact with the central grid cell or are within a preset threshold distance is extracted. Connections are established between each central grid cell and its adjacent grid cells, and basic spatial weights are assigned to these connections according to a distance decay function. The economic attribute difference value of the corresponding grid cells is introduced as a local statistical input and coupled with the basic spatial weights to form weighted connection edges between cells. The weighted connection edges of all grid cells are summarized using a row-column indexing method to construct a spatial adjacency matrix. The spatial... For each row of the adjacency matrix, the neighborhood attribute value sequence of the corresponding central grid cell is extracted. The standardized deviation of the neighborhood attribute value sequence relative to the attribute value of the central grid cell is calculated. Based on the corresponding coupling relationship between the weighted connecting edges and the standardized deviation, the association response value representing the spatial clustering association strength between cells is calculated, and a local spatial deviation vector is established. The local spatial deviation vector is fused element-wise with the weighted connecting edges in the spatial adjacency matrix to generate an association coefficient matrix representing the spatial dependence strength between cells. The association coefficient matrix is ​​hierarchically mapped and spatially reconstructed according to the numerical distribution, and the spatial association relationships of different strength levels are mapped to the corresponding grid positions to form a spatial autocorrelation distribution feature map.

[0026] Preferably, the comprehensive spatial distribution grid field of the port economy is read, and the study area is divided into regular grid cells with a preset fixed spatial step size as the grid side length. Each grid cell is assigned a unique row and column index, denoted as the i-th grid cell. At the same time, the coordinates of the center point of the grid cell and the comprehensive economic intensity value of the port are extracted. Subsequently, a neighborhood window is established with the i-th grid cell as the center, and the radius of the neighborhood window is set to R. The system calculates the Euclidean distance between the center points of other grid cells within the neighborhood window and the central grid cell. Grid cells that have boundary contact with the central grid cell and whose center point distance is no greater than a preset threshold D are identified as adjacent grid cells, forming a set of adjacent grid cells corresponding to the central grid cell. Next, a connection relationship is established between the central grid cell i and its adjacent grid cells j one by one, and a distance decay function is used to calculate the basic spatial weight. Specifically, the basic spatial weight is denoted as... and according to the formula Calculate, where, Let be the distance between the center points of the central grid cell and its adjacent grid cell. When two grid cells are not adjacent, let . =0, then calculate the economic attribute difference value between the two raster cells, and denote it as Furthermore, the economic attribute difference value is coupled with the basic spatial weight to obtain the weighted connection edges between units. ,Right now, Among them, the comprehensive economic intensity value of ports has been normalized to the range of 0 to 1, therefore This is used to characterize the similarity of economic attributes between adjacent grid cells. Therefore, the closer the grid cells are in space and the more similar their economic attributes are, the larger the weighted connection edge value will be.

[0027] Next, according to the row and column indices of the raster cells, the weighted connecting edges between all central raster cells and their adjacent raster cells are written into the corresponding positions in the matrix to construct a spatial adjacency matrix A. In this spatial adjacency matrix, the matrix rows represent central raster cells, the matrix columns represent adjacent raster cells, and the matrix elements... This represents the weighted adjacency relationship between the i-th and j-th grid cells. If they are not adjacent cells, the corresponding matrix element is 0. Then, local spatial statistical calculations are performed on each row vector in the spatial adjacency matrix. For the i-th central grid cell, the comprehensive port economic intensity values ​​of its neighboring grid cells are extracted to form a sequence of neighborhood attribute values. And calculate the standardized deviation of the neighborhood attribute values ​​relative to the attribute value of the central grid cell, denoted as ,in, Let be the standard deviation of the attribute values ​​within the neighborhood window of the i-th central grid cell, when season =0, then add weighted connecting edges Deviation from standardization Perform corresponding coupling and calculate the associated response value. Then, all associated response values ​​corresponding to the same central grid cell are combined to form a local spatial deviation vector. Finally, the local spatial deviation vector is fused element-wise with the weighted connection edges in the spatial adjacency matrix to obtain the correlation coefficient matrix C, where, The correlation coefficient matrix is ​​used to characterize the spatial dependency strength between the i-th and j-th raster units. Based on the numerical distribution of each element in the correlation coefficient matrix, the system classifies the correlation coefficients into high-strength, medium-strength, low-strength, and weak-strength levels. The correlation strength result corresponding to each central raster unit is then remapped to the spatial raster according to its original row and column position, forming a spatial autocorrelation distribution feature map. Through this process, the similarity, differences, and spatial distance influences of the comprehensive economic intensity of a port within its neighborhood can be uniformly quantified, improving the precision and interpretability of identifying spatial agglomeration relationships in the port economy.

[0028] like Figure 2 As shown in the figure, spatial relationships of different intensity levels are presented by graded color bands. The three hot core areas (hot core 1-3) are marked with dashed circles. The contour lines clearly depict the spatial gradient changes in the intensity of the association, intuitively showing the spatial agglomeration pattern and association structure of port economic activities.

[0029] Furthermore, the preset radius and preset threshold of the neighborhood window are adaptively adjusted according to the spatial distribution density of the port economic space integrated distribution grid field. The higher the density, the lower the preset radius and preset threshold are.

[0030] Optionally, when determining the preset radius and preset threshold of the neighborhood window, the effective grid cells in the port economic spatial comprehensive distribution grid field are first used as the statistical objects. Grid cells with a comprehensive intensity value greater than the preset effective intensity value are identified as effective units of economic activity, and an initial statistical window is established with the target grid cell as the center. Subsequently, the number of effective units of economic activity within the initial statistical window is counted, and the ratio of this number to the area of ​​the initial statistical window is used as the spatial distribution density value at the corresponding location of the target grid cell. Then, the neighborhood window radius and the distance threshold between adjacent units are adaptively adjusted based on the spatial distribution density value. Specifically, a baseline neighborhood radius is preset. Reference distance threshold and density adjustment coefficient and in accordance with , Calculate the neighborhood window radius and distance threshold of the i-th grid cell, where, Let represent the spatial distribution density value of the i-th raster cell. A higher spatial distribution density value indicates concentrated port economic activity and close spatial relationships between raster cells. In this case, the neighborhood window radius and distance threshold are automatically reduced, selecting only adjacent raster cells within a closer range for spatial correlation calculation. Conversely, a lower spatial distribution density value indicates sparse port economic activity. Therefore, the neighborhood window radius and distance threshold are increased to include adjacent raster cells within a larger range in the neighborhood analysis. Through this adaptive adjustment, different density regions can obtain a neighborhood range that matches their spatial distribution, avoiding the smoothing of local clustering characteristics caused by excessively large neighborhoods in high-density regions, and preventing the inability to fully identify spatial correlation relationships due to excessively small neighborhoods in low-density regions. This improves the precision and stability of the spatial autocorrelation analysis results.

[0031] like Figure 3 As shown, the effectiveness of the adaptive adjustment mechanism is verified by comparing the neighborhood window settings in high-density and low-density regions. In the left figure (a), the spatial units are densely distributed in the high-density region, and the adaptive neighborhood radius automatically decreases to 5 to ensure the precision of local statistics. In the right figure (b), the spatial units are sparsely distributed in the low-density region, and the adaptive neighborhood radius automatically expands to 15 to ensure sufficient sample size to support statistical reliability. The comparison between the two figures shows that the neighborhood parameter can be dynamically adjusted according to the spatial distribution density, balancing analytical accuracy and statistical stability.

[0032] Hotspot identification and clustering partitioning are performed on the spatial autocorrelation distribution feature map, and the identification partitioning results are established.

[0033] In one embodiment, the association strength value of each grid cell in the spatial autocorrelation distribution feature map is read, and local hotspot scanning is performed on each grid cell sequentially according to its row and column positions. For any target grid cell, the association strength value within its neighborhood is extracted with the target grid cell as the center. The intensity difference between the target grid cell and its neighboring grid cells is calculated, and it is determined whether the target grid cell is a local maximum grid cell within its neighborhood. When the target grid cell meets a preset intensity threshold and is not dominated by higher intensity cells within its spatial neighborhood, the target grid cell is determined as the initial hotspot core cell. Subsequently, using the initial hotspot core cell as the starting point for cluster expansion, the aforementioned association coefficient matrix is ​​invoked to search for adjacent grid cells that have a spatial connection relationship with the hotspot core cell layer by layer, and grid cells whose association strength decreases along the outward expansion direction are merged into the same clustering region. For grid cells already incorporated into a clustered region, the search continues outwards from these cells as new boundary units until the correlation coefficient of the outermost adjacent grid cells falls below the expansion threshold, or until the outermost adjacent grid cells are occupied by other higher-intensity hotspot core units. After expanding all hotspot core units, the boundary grids between adjacent clustered regions are checked to determine whether to retain the partition boundaries between them. Finally, a unique partition number is generated for each spatial clustering partition and mapped back to the spatial location of the study area, forming the identification and partitioning results of the port economic space. This enables the quantitative division of high-value clustered areas, continuous expansion areas, and low-correlation transition areas in the port economy, improving the accuracy of hotspot area identification and cluster boundary determination.

[0034] Furthermore, hotspot identification and clustering are performed on the spatial autocorrelation distribution feature map, and the identification and partitioning results are established, including: Local intensity scanning is performed on each grid cell in the spatial autocorrelation distribution feature map to extract the neighborhood association intensity distribution sequence centered on the target grid cell, and the intensity difference gradient between the central grid cell and its neighboring cells is calculated. Based on the intensity difference gradient, local maxima grid cells are identified, and grid cells that meet a preset intensity threshold and are not dominated by higher intensity cells within the spatial neighborhood are determined as initial hotspot core cells. Using the initial hotspot core cells as seed nodes, neighborhood expansion aggregation is performed according to the association coefficient matrix between grid cells, and spatial connectivity clusters are formed layer by layer according to the decreasing cluster association intensity constraint. Boundary convergence processing is performed on the spatial connectivity clusters. When the cluster association intensity between adjacent clusters is lower than the segmentation threshold, mutually independent spatial cluster partitioning units are formed. All spatial cluster partitioning units are numbered and mapped with spatial identifiers to generate the identification and partitioning results of the port economic space.

[0035] Optionally, read the spatial autocorrelation distribution feature map and record the spatial correlation strength value of each effective raster cell in the map as... The system takes the i-th grid cell as the target grid cell and extracts its surrounding neighboring grid cells according to a preset neighborhood radius, forming a neighborhood association strength distribution sequence. Where j represents the neighboring raster cells within the target raster cell's neighborhood. Subsequently, the intensity difference gradient between the target raster cell and each neighboring cell is calculated, i.e., the correlation intensity value of the target raster cell is... Subtract the correlation strength value of the neighboring unit Divide by the distance between their center points. The intensity difference gradient is obtained. When all intensity difference gradients corresponding to a target raster cell are not less than 0, and the correlation strength value of the target raster cell is greater than a preset intensity threshold, the target raster cell is determined as a local maximum raster cell. If there are no raster cells with higher correlation strength values ​​in the neighborhood of the local maximum raster cell, it is determined as an initial hotspot core cell. After determining the initial hotspot core cells, each initial hotspot core cell is used as a seed node, and the connection relationship corresponding to the seed node in the correlation coefficient matrix is ​​called to perform neighborhood expansion aggregation layer by layer. Specifically, starting from the seed node, the system reads the raster cells directly adjacent to it, determines whether the correlation coefficient between the adjacent raster cell and the current cluster boundary cell is greater than a preset expansion threshold, and determines whether the correlation strength value of the adjacent raster cell is not higher than the correlation strength value of the current boundary cell. When both of the above conditions are met, the adjacent raster cell is merged into the spatial connectivity cluster corresponding to the current hotspot core. For newly incorporated grid cells, the above expansion judgment is repeated as a new boundary cell until the correlation coefficient of the surrounding adjacent grid cells is lower than the preset expansion threshold, or the correlation strength value of the surrounding adjacent grid cells is higher than that of the current boundary cell and does not meet the decreasing constraint. At this point, the expansion in this direction stops. Through layer-by-layer expansion, grid cells with continuous spatial connectivity and decreasing clustering strength around the same hotspot core are grouped into the same spatially connected clustering domain.

[0036] After forming spatially connected clusters, boundary convergence processing is performed on the contact boundaries between adjacent clusters. Specifically, the system extracts adjacent raster cell pairs on the boundaries of two adjacent clusters and calculates the average clustering correlation strength between these pairs based on the correlation coefficient matrix. When the average clustering correlation strength is lower than a preset segmentation threshold, the segmentation boundary between the two clusters is retained, and they are identified as mutually independent spatial clustering partition units. When the average clustering correlation strength is not lower than the preset segmentation threshold, the boundary raster cells between the two clusters are merged into the side with the higher correlation strength, until the average clustering correlation strength on both sides of the boundary meets the segmentation conditions. Finally, a unique number is generated for each partition according to the spatial positional order of the clustering partition units, and the partition number, hotspot core raster position, partition boundary raster set, partition area, and average correlation strength are written into the spatial attribute table. At the same time, the partition number is mapped back to the original raster position, generating the identification partitioning results of the port economic space. Through the above process, areas with high spatial autocorrelation strength and continuous adjacency can be divided into clear port economic clustering partitions, improving the accuracy of hotspot core positioning and cluster boundary identification.

[0037] Furthermore, the preset intensity threshold is determined based on the mean and standard deviation of the correlation intensity of all grid cells in the spatial autocorrelation distribution feature map.

[0038] Optionally, when determining the preset intensity threshold, firstly, the correlation intensity values ​​of all valid raster cells in the spatial autocorrelation distribution feature map are read, and raster cells with no data and those outside the study area boundary are removed, retaining only the valid correlation intensity values ​​used in the spatial autocorrelation calculation. Then, the correlation intensity values ​​of all valid raster cells are summed and divided by the number of valid raster cells to obtain the mean correlation intensity. Simultaneously, the squared difference between the correlation intensity value of each valid raster cell and the mean correlation intensity is calculated, and the square root of the average of all squared differences is taken to obtain the standard deviation of the correlation intensity. Next, the preset intensity threshold is determined by adding the mean and standard deviation; that is, the mean correlation intensity is added to the standard deviation of the correlation intensity by a preset multiple, serving as the intensity judgment benchmark for hotspot core identification, such as 1. When the correlation intensity value of a raster cell is greater than the preset intensity threshold, it indicates that the spatial correlation intensity of the raster cell is significantly higher than the general level within the study area, meeting the criteria for being judged as a hotspot core candidate cell; when its correlation intensity value is not greater than the preset intensity threshold, it is not considered a hotspot core candidate cell for subsequent hotspot expansion. By dynamically determining the preset intensity threshold based on the mean and standard deviation, the threshold can be automatically adjusted according to different port areas, different analysis years, and different spatial distributions of intensity, avoiding the problem of too many or too few hotspots being identified due to a fixed threshold, and improving the adaptability and stability of hotspot core screening.

[0039] The geometric centroid migration trajectory of the cluster center is calculated from the identification and partitioning results of the multi-temporal grid field, and a spatial evolution path sequence is constructed.

[0040] In one embodiment, the multi-temporal port economic spatial integrated distribution raster field and its corresponding identification partition results are read sequentially according to the analysis year, and the raster cells within the same partition are used as the calculation objects for the cluster center. For a spatial clustering partition in any temporal phase, all raster cells contained in the partition are extracted, and the center point coordinates and corresponding clustering association strength values ​​of each raster cell are read. The clustering association strength values ​​are used as weights to perform a weighted average of the center point coordinates of all raster cells in the partition, obtaining the weighted geometric centroid of the spatial clustering partition in that temporal phase. Subsequently, according to the partition number, hotspot core location, and spatial overlap relationship of the partitions, the spatial clustering partitions in two adjacent temporal phases are matched to determine the correspondence between the partitions of the previous and subsequent temporal phases. For the same clustering partition that is successfully matched, its corresponding geometric centroid positions in different years are connected to form the centroid migration trajectory of the clustering partition. Next, the displacement direction, displacement distance, and time interval between the geometric centroids of adjacent time phases are calculated. The migration rate is obtained by dividing the displacement distance by the time interval. The displacement results between each time phase are arranged in chronological order to construct a spatial evolution path sequence. This spatial evolution path sequence includes at least the time phase identifier, partition number, geometric centroid coordinates, migration direction, migration distance, and migration rate, and is used to characterize the spatial movement process of the port economic agglomeration center at different times. Through the above process, the spatial distribution of the port economy can be transformed from a static partition result into a continuous temporal evolution trajectory, thereby improving the accuracy of the analysis of the port economic agglomeration direction, diffusion range, and spatial change trend.

[0041] Furthermore, constructing a spatial evolution path sequence includes: The identification partitioning results of the multi-temporal grid field at each time are extracted respectively, and the weighted geometric centroid of each partition unit is calculated based on the grid position and the clustering correlation strength. The weighted geometric centroids of the same partition unit in continuous temporal phases are matched to establish the geometric centroid migration trajectory. Based on the geometric centroid migration trajectory, the spatial displacement vector of adjacent temporal phases is calculated, and the spatial evolution path sequence is constructed according to the time series.

[0042] Preferably, the multi-phase raster fields are first read in chronological order, for example, the first phase is read sequentially. , No. , No. The corresponding port economic spatial comprehensive distribution raster field and its identification and partitioning results for each year are maintained, ensuring that the raster fields of each time phase have the same coordinate system, raster size, and row and column numbering. For the m-th partition cell at any time phase t, all raster cells contained in that partition cell are extracted, and the center point coordinates of each raster cell are read. and the corresponding clustering correlation strength value The clustering correlation strength value is used as the weight to calculate the weighted geometric centroid of the partition unit, specifically: , Where n is the number of raster cells in the partition unit. Let be the weighted geometric centroid of the m-th partition unit at time phase t. When the sum of the clustering correlation strength values ​​of all raster units within a partition unit is 0, the arithmetic mean of the coordinates of all raster center points within that partition unit is used as the geometric centroid. After calculating the weighted geometric centroid of each partition unit at each time phase, partition units at consecutive time phases are matched. That is, taking the partition unit at the previous time phase t as the benchmark, the spatial overlap area ratio between it and each partition unit at the next time phase t+1 is calculated, and the partition unit at the next time phase with the largest spatial overlap area ratio is selected as the candidate matching object. When the spatial overlap area ratio is greater than a preset matching threshold, the two partition units are identified as the corresponding results of the same clustering partition at consecutive time phases. For successfully matched partition units, their weighted geometric centroids at consecutive time phases are connected sequentially to form a geometric centroid migration trajectory.

[0043] Subsequently, the weighted geometric centroid of the same partition unit at time phases t and t+1. and Calculate the spatial displacement vector: The process further calculates the displacement distance, migration direction angle, and migration rate. The displacement distance is calculated by summing the squares of the differences in the lateral and longitudinal coordinates of the weighted geometric centroids of adjacent time phases and then taking the square root. The migration direction angle is calculated by using the arctangent function to calculate the angle between the displacement vector and the due east direction, with the lateral coordinate difference as the x-axis component and the longitudinal coordinate difference as the y-axis component. The migration rate is calculated by dividing the displacement distance by the time interval between two adjacent time phases. Finally, according to the chronological order of the time phases, the partition number, starting time phase, ending time phase, starting centroid coordinates, ending centroid coordinates, spatial displacement vector, displacement distance, migration direction angle, and migration rate are written into the path sequence table to generate a spatial evolution path sequence. Through this process, the multi-time phase identification partitioning results can be transformed into quantifiable spatial migration trajectories, improving the accuracy of judging the movement direction and evolution trend of port economic agglomeration centers.

[0044] Furthermore, constructing a spatial evolution path sequence includes: The direction angle and magnitude of the spatial displacement vector corresponding to each time step in the spatial evolution path sequence are extracted, and time-series curves of the main migration direction and migration rate of the agglomeration center are statistically generated. Based on the time-series curves of the main migration direction and migration rate, the agglomeration strengthening stage, diffusion spillover stage, and multi-center fission stage of the port economic spatial distribution are identified. The identification partition results of the corresponding time phases of each evolution stage are extracted respectively. After calculating the standard deviation ellipse, the major axis direction of the ellipse is extracted as the main spatial diffusion direction, and the ellipse area change rate is calculated. The ellipse area change rate characterizes the changing trend of the port economic radiation range. The evolution law and dynamic response characteristics of the port economic spatial distribution are obtained by combining the main migration direction, the main spatial diffusion direction, and the changing trend.

[0045] Preferably, the system first reads the constructed spatial evolution path sequence and extracts the spatial displacement vector corresponding to each adjacent time period. For the spatial displacement vector of the m-th agglomeration zone between time period t and time period t+1, its lateral and longitudinal displacement components are read, and the orientation angle and magnitude of the displacement vector are calculated. The orientation angle represents the migration direction of the agglomeration center from the previous time period to the next time period; the magnitude represents the migration distance of the agglomeration center within that time interval. The system weights and statistically calculates the orientation angles of each zone unit within the same time period according to the agglomeration association strength or zone area of ​​the corresponding zone to obtain the main migration direction of the agglomeration center in that time period. At the same time, the displacement magnitude of each zone unit is divided by the time interval between adjacent time periods to obtain the migration rate, and a migration rate time series curve is generated in chronological order. Subsequently, the evolution stage of the port's economic spatial distribution is identified based on the main migration direction and migration rate time series curve. Specifically, when the migration rate of clustering centers is low, the area of ​​spatial clustering zones changes little, and the number of highly correlated grids increases within a continuous time phase, this period is identified as the clustering reinforcement stage; when the migration rate of clustering centers increases, the main migration direction remains relatively stable, and the boundaries of clustering zones expand outward, this period is identified as the diffusion and spillover stage; when two or more independent hotspot core zones appear within the same time phase, and the differences in the main migration directions of different zones increase, this period is identified as the multicenter fission stage. Thus, the spatial variation process of consecutive years can be divided into stages with different spatial evolution characteristics.

[0046] Subsequently, the identification and partitioning results corresponding to each evolutionary stage are extracted, and the coordinates of the center point of the grid cells within each partition and their clustering correlation strength are used as inputs to calculate the standard deviation ellipse. Specifically, the weighted average center of the grid cells within the partition is first calculated, then the lateral and longitudinal deviations of each grid cell relative to the weighted average center are calculated, and the major axis, minor axis, and rotation angle of the ellipse are determined based on the deviation covariance relationship. The system uses the major axis direction of the standard deviation ellipse as the main direction of spatial diffusion for that evolutionary stage, and calculates the ellipse area based on the major and minor axis lengths. Then, the difference between the ellipse area of ​​the later time phase and the ellipse area of ​​the previous time phase is divided by the ellipse area of ​​the previous time phase to obtain the ellipse area change rate. When this ellipse area change rate is greater than 0, it indicates that the port's economic radiation range is expanding; when it is less than 0, it indicates that the port's economic radiation range is contracting; and when it is close to 0, it indicates that the radiation range is relatively stable. Then, the main direction of agglomeration center migration, the time series curve of migration rate, the direction of the major axis of the standard deviation ellipse, and the rate of change of ellipse area are jointly interpreted. That is, the angle difference between the main direction of agglomeration center migration and the direction of the major axis of the standard deviation ellipse in each time phase within the same evolution stage is calculated to obtain the direction consistency index. When the angle difference between the two is less than the preset angle threshold, it is determined that the migration direction of the port economic agglomeration center is consistent with the spatial diffusion direction, indicating that the port economy is expanding outward along the same dominant direction. When the angle difference between the two is greater than the preset angle threshold, it is determined that the migration direction of the agglomeration center is inconsistent with the spatial diffusion direction, indicating that the spatial distribution of the port economy has the characteristics of offset expansion or multi-directional dispersion. Next, the system reads the trend of migration rate time-series curves, calculates the increase or decrease in migration rate between adjacent time phases, and combines this with the ellipse area change rate to determine the radiation range change status. Generally, when the migration rate is low and the ellipse area change rate is close to zero, while the correlation strength within high-intensity agglomeration zones increases, it is identified as agglomeration strengthening characteristic; when the migration rate increases and the ellipse area change rate is positive, and the main migration direction is consistent with the major axis of the ellipse, it is identified as diffusion spillover characteristic; when the difference in migration rate between different zones increases, the ellipse area change rate is positive, and multiple independent hotspot cores appear, it is identified as multi-center fission characteristic; when the migration rate decreases and the ellipse area change rate is negative, it is identified as agglomeration contraction characteristic. Based on the above determination results, the system generates corresponding spatial evolution type, dominant migration direction, dominant diffusion direction, radiation range change trend, and stage characteristic descriptions for each evolution stage, thereby obtaining the evolutionary law and dynamic response characteristics of the port's economic spatial distribution. Through the above process, it is possible to simultaneously characterize the changes in the direction of movement, speed of movement, spatial diffusion direction, and radiation range of port economic agglomeration centers, thereby improving the quantitative analysis capability of the spatial agglomeration, spillover, and multi-center evolution process of port economy.

[0047] In summary, the embodiments of this application have at least the following technical effects: First, multi-source heterogeneous data is acquired interactively. This data includes port economic value-added data, port-adjacent enterprise distribution data, transportation network data, and land use data. After coordinate unification and scale normalization, this multi-source heterogeneous data forms a standardized spatial element dataset. Next, the standardized spatial element dataset is rasterized to construct a multi-layer spatial representation system with unified resolution. Weight assignment and overlay fusion of each element layer are performed within this system to generate a comprehensive port economic spatial distribution raster field. Then, a spatial adjacency matrix is ​​constructed based on this raster field. Within this matrix, the clustering correlation strength between spatial units is calculated using local spatial statistical indicators, generating a spatial autocorrelation distribution feature map. Next, hotspot identification and clustering partitioning are performed on the spatial autocorrelation distribution feature map, establishing the identification partitioning results. Finally, the geometric centroid migration trajectory of the clustering centers is calculated based on the identification partitioning results of the multi-temporal raster field, constructing a spatial evolution path sequence. This invention addresses the technical challenge of accurately identifying the spatial distribution characteristics of port economies and quantitatively analyzing their clustering and evolution trends due to inconsistencies in coordinate benchmarks, spatial scales, and temporal scales in multi-source heterogeneous data related to port economies. It achieves the technical effect of improving the accuracy of identifying the spatial distribution of port economies and the accuracy of analyzing spatial clustering and evolution trends by standardizing spatial representation, layer overlay and fusion, spatial autocorrelation analysis, and multi-temporal evolution modeling of multi-source heterogeneous data.

[0048] Example 2, based on the same inventive concept as the GIS-based port economic spatial distribution analysis method in the previous examples, such as... Figure 4 As shown, this application provides a GIS-based port economic spatial distribution analysis system, wherein the system includes: Data Standardization Module 11: Interactively acquires multi-source heterogeneous data, including port economic added value data, port-adjacent enterprise distribution data, transportation network and land use data. After coordinate unification and scale normalization, the multi-source heterogeneous data forms a standardized spatial element dataset. Element Fusion Module 12: Rasterizes the standardized spatial element dataset, constructs a multi-layer spatial expression system with unified resolution, assigns weights to each element layer and performs overlay fusion in the multi-layer spatial expression system, generating a comprehensive distribution raster field of port economic space. Association Analysis Module 13: Constructs a spatial adjacency matrix based on the comprehensive distribution raster field of port economic space, calculates the clustering correlation strength between spatial units based on local spatial statistical indicators in the spatial adjacency matrix, and generates a spatial autocorrelation distribution feature map. Clustering Partitioning Module 14: Identifies hotspots and clusters in the spatial autocorrelation distribution feature map, and establishes identification and partitioning results. Evolution Analysis Module 15: Calculates the geometric centroid migration trajectory of the clustering center based on the identification and partitioning results of the multi-temporal raster field, and constructs a spatial evolution path sequence.

[0049] Furthermore, the data standardization module 11 is used to perform the following methods: Spatial geographic registration and projection transformation are performed on the port economic value-added data and the port-adjacent enterprise distribution data to place the port economic value-added data, port-adjacent enterprise distribution data, traffic routes and land use data in the same geographic coordinate system; the data of each element are resampled and time sliced ​​according to a unified spatial resolution and raster size to eliminate scale differences and spatiotemporal misalignment, and a standardized spatial element dataset is established.

[0050] Furthermore, the element fusion module 12 is used to perform the following method: Normalization was performed on the port economic value-added layer, port-adjacent enterprise distribution layer, transportation network layer, and land use layer, transforming the elements of each layer into rasterized numerical expressions with the same dimension. Based on the driving contribution of each layer to port economic development, weight coefficients for economic value-added, industry density, transportation accessibility, and land use suitability were initialized. The entropy method was used to calculate the dispersion of raster values ​​in each layer, and information entropy weight correction factors for each layer were established. The initialized weight coefficients were dynamically corrected using the information entropy weight correction factors, and weight combinations were established to complete the weight assignment. Based on the weight assignment, a pixel-by-pixel weighted overlay operation was performed on each element layer, and the operation results were subjected to nonlinear stretching and extreme value suppression processing to generate a comprehensive spatial distribution raster field of port economy.

[0051] Furthermore, the correlation analysis module 13 is used to perform the following methods: The port economic spatial distribution grid field is divided into regular grid cells with a fixed spatial step size. A neighborhood window with a preset radius is established with each grid cell as the center. The set of adjacent grid cells within the neighborhood window that have boundary contact with the central grid cell or are less than a preset threshold distance is extracted. A connection relationship is established between each central grid cell and its adjacent grid cells one by one, and a basic spatial weight is assigned to the connection relationship according to the distance decay function. The economic attribute difference value of the corresponding grid cell is introduced as a local statistical input, and coupled with the basic spatial weight, a weighted connection edge between the cells is formed. The weighted connection edges of all grid cells are summarized in a row-column index manner to construct a spatial adjacency matrix. For each row of the adjacency matrix, the neighborhood attribute value sequence of the corresponding central grid cell is extracted. The standardized deviation of the neighborhood attribute value sequence relative to the attribute value of the central grid cell is calculated. Based on the corresponding coupling relationship between the weighted connecting edges and the standardized deviation, the association response value representing the spatial clustering association strength between cells is calculated, and a local spatial deviation vector is established. The local spatial deviation vector is fused element-wise with the weighted connecting edges in the spatial adjacency matrix to generate an association coefficient matrix representing the spatial dependence strength between cells. The association coefficient matrix is ​​hierarchically mapped and spatially reconstructed according to the numerical distribution, and the spatial association relationships of different strength levels are mapped to the corresponding grid positions to form a spatial autocorrelation distribution feature map.

[0052] Furthermore, the correlation analysis module 13 is used to perform the following methods: The preset radius and preset threshold of the neighborhood window are adaptively adjusted according to the spatial distribution density of the port economic space integrated distribution grid field. The higher the density, the lower the preset radius and preset threshold.

[0053] Furthermore, the aggregation partitioning module 14 is used to perform the following method: Local intensity scanning is performed on each grid cell in the spatial autocorrelation distribution feature map to extract the neighborhood association intensity distribution sequence centered on the target grid cell, and the intensity difference gradient between the central grid cell and its neighboring cells is calculated. Based on the intensity difference gradient, local maxima grid cells are identified, and grid cells that meet a preset intensity threshold and are not dominated by higher intensity cells within the spatial neighborhood are determined as initial hotspot core cells. Using the initial hotspot core cells as seed nodes, neighborhood expansion aggregation is performed according to the association coefficient matrix between grid cells, and spatial connectivity clusters are formed layer by layer according to the decreasing cluster association intensity constraint. Boundary convergence processing is performed on the spatial connectivity clusters. When the cluster association intensity between adjacent clusters is lower than the segmentation threshold, mutually independent spatial cluster partitioning units are formed. All spatial cluster partitioning units are numbered and mapped with spatial identifiers to generate the identification and partitioning results of the port economic space.

[0054] Furthermore, the aggregation partitioning module 14 is used to perform the following method: The preset intensity threshold is determined based on the mean and standard deviation of the correlation intensity of all grid cells in the spatial autocorrelation distribution feature map.

[0055] Furthermore, the evolutionary analysis module 15 is used to perform the following methods: The identification partitioning results of the multi-temporal grid field at each time are extracted respectively, and the weighted geometric centroid of each partition unit is calculated based on the grid position and the clustering correlation strength. The weighted geometric centroids of the same partition unit in continuous temporal phases are matched to establish the geometric centroid migration trajectory. Based on the geometric centroid migration trajectory, the spatial displacement vector of adjacent temporal phases is calculated, and the spatial evolution path sequence is constructed according to the time series.

[0056] Furthermore, the evolutionary analysis module 15 is used to perform the following methods: The direction angle and magnitude of the spatial displacement vector corresponding to each time step in the spatial evolution path sequence are extracted, and time-series curves of the main migration direction and migration rate of the agglomeration center are statistically generated. Based on the time-series curves of the main migration direction and migration rate, the agglomeration strengthening stage, diffusion spillover stage, and multi-center fission stage of the port economic spatial distribution are identified. The identification partition results of the corresponding time phases of each evolution stage are extracted respectively. After calculating the standard deviation ellipse, the major axis direction of the ellipse is extracted as the main spatial diffusion direction, and the ellipse area change rate is calculated. The ellipse area change rate characterizes the changing trend of the port economic radiation range. The evolution law and dynamic response characteristics of the port economic spatial distribution are obtained by combining the main migration direction, the main spatial diffusion direction, and the changing trend.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A GIS-based method for analyzing the spatial distribution of port economic activity, characterized in that, The method includes: Interactive acquisition of multi-source heterogeneous data, including port economic added value data, port-adjacent enterprise distribution data, transportation network and land use data, and the multi-source heterogeneous data are converted into a standardized spatial element dataset after coordinate unification and scale normalization. The standardized spatial element dataset is rasterized to construct a multi-layer spatial representation system with a unified resolution. In the multi-layer spatial representation system, the weights of each element layer are assigned and superimposed and fused to generate a comprehensive distribution raster field of port economic space. A spatial adjacency matrix is ​​constructed based on the comprehensive spatial distribution grid field of port economy. The clustering correlation strength between spatial units is calculated based on local spatial statistical indicators in the spatial adjacency matrix to generate a spatial autocorrelation distribution feature map. Hotspot identification and clustering partitioning of spatial autocorrelation distribution feature maps were performed, and the identification and partitioning results were established. The geometric centroid migration trajectory of the cluster center is calculated from the identification and partitioning results of the multi-temporal grid field, and a spatial evolution path sequence is constructed.

2. The GIS-based port economic spatial distribution analysis method as described in claim 1, characterized in that, Generate a spatial autocorrelation distribution feature map, including: The port economic spatial distribution grid field is divided into regular grid units according to a fixed spatial step size; Establish a neighborhood window with a preset radius centered on each grid cell, and extract the set of adjacent grid cells within the neighborhood window that have boundary contact with the central grid cell or are less than a preset threshold. For each central grid cell, establish a connection relationship with each of its adjacent grid cells, and assign a basic spatial weight to the connection relationship according to the distance decay function; The economic attribute difference value of the corresponding grid cell is introduced as a local statistical input, and coupled calculation is performed in combination with the basic spatial weight to form a weighted connection edge between cells. Summarize the weighted connection edges of all grid cells by row-column index to construct a spatial adjacency matrix; For each row vector in the spatial adjacency matrix, extract the neighborhood attribute value sequence corresponding to the central grid cell, calculate the standardized deviation of the neighborhood attribute value sequence relative to the attribute value of the central grid cell, calculate the association response value characterizing the spatial clustering association strength between cells based on the corresponding coupling relationship between the weighted connection edge and the standardized deviation, and establish a local spatial deviation vector. The local spatial deviation vector is fused with the weighted connection edges in the spatial adjacency matrix on an element-by-element basis to generate a correlation coefficient matrix that represents the spatial dependence strength between units. The correlation coefficient matrix is ​​hierarchically mapped and spatially reconstructed according to its numerical distribution, and the spatial correlation relationships of different strength levels are mapped to the corresponding grid positions to form a spatial autocorrelation distribution feature map.

3. The GIS-based spatial distribution analysis method for port economies as described in claim 2, characterized in that, Hotspot identification and clustering partitioning of spatial autocorrelation distribution feature maps were performed, and the identification and partitioning results were established, including: Local intensity scanning is performed on each grid cell in the spatial autocorrelation distribution feature map to extract the neighborhood correlation intensity distribution sequence centered on the target grid cell, and the intensity difference gradient between the central grid cell and its neighboring cells is calculated. Based on the intensity difference gradient identification of local maxima grid cells, grid cells that meet the preset intensity threshold and are not dominated by higher intensity cells in the spatial neighborhood are determined as initial hot spot core cells. Using the initial hotspot core unit as the seed node, neighborhood expansion aggregation is performed based on the correlation coefficient matrix between grid units, and spatially connected cluster domains are formed by expanding layer by layer according to the constraint of decreasing cluster correlation strength. Boundary convergence processing is performed on the spatially connected cluster domains. When the cluster association strength between adjacent cluster domains is lower than the segmentation threshold, mutually independent spatial cluster partition units are formed. All spatial clustering and partitioning units are numbered and mapped with spatial identifiers to generate identification and partitioning results for the port economic space.

4. The GIS-based spatial distribution analysis method for port economies as described in claim 3, characterized in that, The preset intensity threshold is determined based on the mean and standard deviation of the correlation intensity of all grid cells in the spatial autocorrelation distribution feature map.

5. The GIS-based spatial distribution analysis method for port economies as described in claim 1, characterized in that, Constructing a sequence of spatial evolutionary paths, including: The identification partitioning results corresponding to the multi-temporal grid field at each time are extracted respectively, and the weighted geometric centroid of each partitioning unit is calculated based on the grid position and the clustering correlation strength; The weighted geometric centroids of the same partition unit in continuous time phases are matched to establish the geometric centroid migration trajectory; Based on the geometric center of gravity migration trajectory, calculate the spatial displacement vector of adjacent time phases, and construct a spatial evolution path sequence according to the time series.

6. The GIS-based port economic spatial distribution analysis method as described in claim 1, characterized in that, The multi-source heterogeneous data, after coordinate unification and scale normalization, forms a standardized spatial element dataset, including: Spatial geographic registration and projection transformation are performed on the port economic value-added data and the port-adjacent enterprise distribution data so that the port economic value-added data, the port-adjacent enterprise distribution data, and the traffic routes and land use data are in the same geographic coordinate system. By resampling and aligning the data of each element with time slices according to a unified spatial resolution and raster size, scale differences and spatiotemporal misalignment are eliminated, and a standardized spatial element dataset is established.

7. The GIS-based port economic spatial distribution analysis method as described in claim 1, characterized in that, Generate a comprehensive spatial distribution grid field for the port economy, including: The port economic value added layer, port-adjacent enterprise distribution layer, transportation network layer, and land use layer were normalized respectively, and the elements of each layer were converted into rasterized numerical expressions with the same dimension. Based on the contribution of each layer to the development of the port economy, the weight coefficients of economic value added, industrial density, transportation accessibility and land use suitability are initialized. The entropy method is used to calculate the dispersion of raster values ​​in each layer, establish the information entropy weight correction factor for each layer, use the information entropy weight correction factor to dynamically correct the initial weight coefficients, and establish weight combinations to complete the weight assignment. Based on the weight assignment, a pixel-by-pixel weighted overlay operation is performed on each feature layer. The operation results are then subjected to nonlinear stretching and extreme value suppression to generate a comprehensive distribution raster field of port economic space.

8. The GIS-based spatial distribution analysis method for port economies as described in claim 2, characterized in that, The preset radius and preset threshold of the neighborhood window are adaptively adjusted according to the spatial distribution density of the port economic space integrated distribution grid field. The higher the density, the lower the preset radius and preset threshold.

9. The GIS-based spatial distribution analysis method for port economies as described in claim 1, characterized in that, Constructing a sequence of spatial evolutionary paths, including: Extract the orientation angle and magnitude of the spatial displacement vector corresponding to each time step in the spatial evolution path sequence, and statistically generate the time series curves of the main migration direction and migration rate of the agglomeration center; Based on the time-series curves of the main migration direction and migration rate, the agglomeration and intensification stage, diffusion and spillover stage, and multi-center fission stage of the port's economic spatial distribution can be identified. The identification and partitioning results of the corresponding time phases of each evolution stage are extracted respectively. After calculating the standard deviation ellipse, the major axis direction of the ellipse is extracted as the main direction of spatial diffusion, and the ellipse area change rate is calculated. The ellipse area change rate characterizes the changing trend of the port's economic radiation range. By combining the main migration direction, the main spatial diffusion direction, and the changing trend, we can obtain the evolution law and dynamic response characteristics of the spatial distribution of port economy.

10. A GIS-based port economic spatial distribution analysis system, characterized in that, The system is used to implement the GIS-based port economic spatial distribution analysis method according to any one of claims 1-9, the system comprising: Data standardization module: interactively acquires multi-source heterogeneous data, including port economic added value data, port-adjacent enterprise distribution data, transportation network and land use data. After coordinate unification and scale normalization, the multi-source heterogeneous data forms a standardized spatial element dataset. Element fusion module: The standardized spatial element dataset is rasterized to construct a multi-layer spatial expression system with unified resolution. In the multi-layer spatial expression system, the weights of each element layer are assigned and superimposed and fused to generate a comprehensive distribution raster field of port economic space. The correlation analysis module constructs a spatial adjacency matrix based on the comprehensive spatial distribution grid field of port economy. In the spatial adjacency matrix, the clustering correlation strength between spatial units is calculated based on local spatial statistical indicators, and a spatial autocorrelation distribution feature map is generated. Clustering and partitioning module: performs hotspot identification and clustering and partitioning on the spatial autocorrelation distribution feature map, and establishes the identification and partitioning results; Evolution Analysis Module: Calculates the geometric centroid migration trajectory of the clustering center based on the identification and partitioning results of the multi-temporal grid field, and constructs a spatial evolution path sequence.

Images