A method and system for dynamic aggregation of spatial features based on multiple constraints

Abstract

The application provides a multi-constraint-based spatial element dynamic aggregation method and system, and belongs to the technical field of spatial data processing. The method comprises the following steps: acquiring shared bicycle parking point spatial element data set and multi-dimensional constraint conditions; performing spatial profiling on a target research area based on spatial division constraints, generating a plurality of initial aggregation service range units, and mapping elements in the spatial element data set into the initial aggregation service range units, and calculating statistical indexes of the initial aggregation service range units; constructing a graph structure with the initial aggregation service range units as nodes, and the spatial adjacency relationship and semantic similarity between the units as edges; constructing a graph structure aggregation model, including a target function with the minimum variance of the statistical indexes between the initial aggregation service range units as the target, and a constraint condition based on the integrity of the spatial division constraints; and solving the graph structure aggregation model by using a heuristic search algorithm or a graph partitioning algorithm to generate a dynamic aggregation service range result satisfying the statistical balance constraint.

Description

Technical Field

[0001] This application relates to the technical field of spatial data processing, and in particular to a method and system for dynamic aggregation of spatial elements based on multiple constraints. Background Technology

[0002] As the scale of shared bicycle operations continues to expand, the number of parking spots in urban areas is large and scattered. Existing methods relying on fixed administrative divisions or manual experience for allocation cannot adapt to real-time changes in riding demand, leading to problems such as mismatch between regional capacity and demand, and imbalances in management units. Existing spatial aggregation methods often rely on single spatial constraints, failing to comprehensively consider multiple dimensions such as land use function, riding interaction, and supply-demand balance, resulting in a disconnect between the allocation results and actual travel patterns. Graph-based partitioning methods rely solely on spatial adjacency relationships, failing to incorporate semantic similarity and business statistical features into edge weight construction, resulting in insufficient expression of regional correlations. Existing aggregation models lack optimization mechanisms aimed at statistical equilibrium, and in iterative solutions, they cannot balance the integrity of spatial boundaries with the balance of regional indicators, leading to situations where some areas are overloaded and others are idle.

[0003] Therefore, there is an urgent need for a method and system for dynamic aggregation of spatial elements based on multiple constraints. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method and system for dynamic aggregation of spatial elements based on multiple constraints.

[0005] A first aspect of this application provides a method for dynamic aggregation of spatial features based on multiple constraints, including: Acquire a spatial element dataset of shared bicycle parking spots and user-defined multidimensional constraints; the multidimensional constraints include at least spatial partitioning constraints and statistical equilibrium constraints. Based on the aforementioned spatial partitioning constraints, the target research area is spatially divided to generate several initial aggregated service range units with basic attributes. The features in the spatial feature dataset are mapped to the initial aggregation service range units, and the statistical indicators of each initial aggregation service range unit are calculated. Construct a graph structure with the initial aggregation service range units as nodes and the spatial adjacency and semantic similarity between units as edges; Based on the graph structure, a graph structure aggregation model is constructed. The graph structure aggregation model includes an objective function that aims to minimize the variance of statistical indicators among the initial aggregation service scope units, and constraints based on maintaining the integrity of the spatial partitioning constraints. The graph structure aggregation model is iteratively solved and partitioned using a heuristic search algorithm or a graph partitioning algorithm. Under the spatial partitioning constraints, adjacent initial aggregation service range units are merged or reorganized to generate a dynamic aggregation service range result that satisfies the statistical equilibrium constraints.

[0006] A second aspect of this application provides a spatial feature dynamic aggregation system based on multiple constraints, comprising: The data acquisition module is used to acquire a spatial element dataset of shared bicycle parking points and user-defined multidimensional constraints; the multidimensional constraints include at least spatial partitioning constraints and statistical equilibrium constraints. The spatial partitioning module is used to spatially partition the target research area based on the spatial partitioning constraints, and generate several initial aggregation service range units with basic attributes. The indicator calculation module is used to map the features in the spatial feature dataset to the initial aggregation service range units and calculate the statistical indicators of each initial aggregation service range unit. The graph construction module is used to construct a graph structure with the initial aggregation service range units as nodes and the spatial adjacency relationship and semantic similarity between units as edges; The model building module is used to build a graph structure aggregation model based on the graph structure. The graph structure aggregation model includes an objective function aimed at minimizing the variance of statistical indicators among the initial aggregation service range units, and constraints based on maintaining the integrity of the spatial partitioning constraints. The optimization solution module is used to iteratively solve and partition the graph structure aggregation model using a heuristic search algorithm or a graph partitioning algorithm. Under the spatial partitioning constraints, it merges or reorganizes adjacent initial aggregation service range units to generate dynamic aggregation service range results that satisfy the statistical equilibrium constraints.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for dynamic aggregation of spatial features based on multiple constraints.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for dynamic aggregation of spatial features based on multiple constraints.

[0009] The beneficial effects of the spatial element dynamic aggregation method and system based on multiple constraints provided in this application are as follows: This application obtains a dataset of spatial elements of shared bicycle parking points and multi-dimensional constraints, generates initial aggregation service range units based on spatial partitioning constraints, maps elements and calculates statistical indicators, constructs a graph structure with units as nodes and adjacency relationships and semantic similarity as edges, establishes an aggregation model with variance minimization as the objective and spatial constraint integrity as the constraint, and finally iteratively solves and merges and recombines units through heuristic or graph partitioning algorithms to generate a dynamic aggregation service range that satisfies statistical equilibrium. This can achieve balanced division of service areas while ensuring compliance with spatial boundaries, improve the rationality, balance and practicality of shared bicycle service range division, and thus improve the efficiency of regional scheduling and management. Attached Figure Description

[0010] Figure 1 A flowchart illustrating a method for dynamic aggregation of spatial elements based on multiple constraints, provided in an embodiment of this application; Figure 2 A structural block diagram of a spatial element dynamic aggregation system based on multiple constraints provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0012] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1-3 The following is an explanation using specific examples.

[0013] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for dynamic aggregation of spatial features based on multiple constraints, provided in an embodiment of this application. The method includes: S101: Obtain the spatial element dataset of shared bicycle parking points and user-defined multidimensional constraints; the multidimensional constraints should include at least spatial partitioning constraints and statistical equilibrium constraints.

[0014] In this embodiment, shared bicycle parking spots are fixed locations within an urban area used for the centralized parking, management, and dispatching of shared bicycles. These include electronically fenced parking areas, recommended parking spots, and designated P-points, possessing attributes such as spatial location, parking capacity, and the area they belong to. The spatial element dataset is a structured data set centered on spatial coordinates, including geometric, attribute, and business information of shared bicycle parking spots; it serves as the foundational data for spatial aggregation. User definitions are rules, parameters, and constraints set by operations personnel, planners, or system administrators based on management needs, dispatching requirements, and planning objectives.

[0015] In this embodiment, multidimensional constraints are a set of multidimensional restriction rules that must be satisfied simultaneously during the dynamic aggregation of spatial elements. These rules ensure that the aggregation results meet requirements in terms of spatial regularity, statistical balance, and business rationality. Spatial partitioning constraints are rigid or semi-rigid constraints on the boundaries of geographic space partitioning, unit morphology, and zoning range, used to limit the generation rules and spatial boundaries of the initial units. Statistical balance constraints are requirements for the balance of statistical indicators among the aggregated service area units, ensuring that the differences in indicators such as the number of parking points, capacity, and demand among units are within a reasonable range, achieving supply and demand equilibrium.

[0016] S102: Based on spatial partitioning constraints, the target research area is spatially divided to generate several initial aggregated service range units with basic attributes.

[0017] In this embodiment, spatial partitioning constraints are user-defined constraints used to limit the type and rules of spatial partitioning boundaries. These constraints include at least one of rule grid constraints, administrative division constraints, and natural street block constraints, ensuring that the boundaries of the aggregation units meet management or planning requirements. The target study area is the urban geographical scope targeted by this dynamic aggregation of shared bicycle parking space elements; it is a closed area with clearly defined geographical boundaries. Spatial partitioning is the process of dividing the continuous geographical space of the target study area into several non-overlapping, fully covered basic spatial units according to spatial partitioning constraints.

[0018] In this embodiment, the initial aggregation service area unit is a basic spatial unit formed after spatial partitioning but before aggregation and reorganization. It is the smallest basic unit for subsequent map construction, index calculation, and iterative aggregation. The basic attributes are the geometric and identification attributes inherent to the initial aggregation service area unit, including unit number, spatial range, boundary coordinates, administrative division, and land use function type.

[0019] S103: Map the features in the spatial feature dataset to the initial aggregation service scope units, and calculate the statistical indicators for each initial aggregation service scope unit.

[0020] In this embodiment, the spatial element dataset is a structured data collection organized around spatial location. It includes geometric information (latitude and longitude, location range), attribute information (parking point ID, capacity), and business information (historical borrowing and returning data) of shared bicycle parking points, serving as the fundamental data source for element mapping and indicator calculation. An element specifically refers to each independent data record in the spatial element dataset, i.e., a single shared bicycle parking point. Each element includes a unique identifier, spatial coordinates, and related attribute information, and is the smallest data unit for mapping and statistics. Mapping involves spatial analysis operations such as spatial indexing and spatial joining to associate and match a single shared bicycle parking point (element) with the initial aggregated service area unit it falls into, clarifying the unit to which each element belongs and realizing the point-to-unit correspondence.

[0021] In this embodiment, the initial aggregation service range unit is a basic spatial unit with basic attributes (unit number, spatial range, etc.) generated after spatial partitioning, and serves as the calculation carrier for statistical indicators. The statistical indicators are quantitative values ​​obtained by statistically analyzing, accumulating, and calculating all shared bicycle parking points falling within the initial aggregation service range unit. These values ​​are used to characterize the distribution, supply and demand, and operational characteristics of shared bicycles within the unit, providing data support for subsequent aggregation optimization.

[0022] S104: Construct a graph structure with the initial aggregation service scope units as nodes and the spatial adjacency and semantic similarity between units as edges.

[0023] In this embodiment, the construction process involves identifying spatial relationships, calculating semantic similarity, and integrating the associations of nodes and edges to build an undirected weighted graph that conforms to the definition of a mathematical graph and adapts to the spatial aggregation requirements of shared bicycle parking points. This process transforms spatial units and their inter-unit relationships into a computationally achievable model. The initial aggregation service range unit is a basic spatial unit generated through spatial partitioning, with completed mapping of shared bicycle parking point elements and calculation of statistical indicators (capacity demand ratio, etc.). It serves as the carrier of attribute information in the graph structure and acts as a basic node. A node is a constituent unit of the graph structure; in this embodiment, it refers to a single initial aggregation service range unit. Each node is bound to a unique identifier (unit number), its own basic attributes (spatial range, boundary coordinates), and calculated statistical indicators, used for subsequent edge weight calculations and model solving.

[0024] In this embodiment, the spatial adjacency relationship between units refers to the geographical proximity of two initial aggregated service scope units. Specifically, it means that the geometric boundaries of the two units share common edges and vertices, which form the basic spatial basis for constructing the graph structure edges. Semantic similarity is a 0-1 range quantitative value obtained by weighting the supply-demand difference (capacity demand ratio difference), land use function correlation, and historical borrowing and returning interaction intensity of the two initial aggregated service scope units. It is used to characterize the degree of similarity between units in terms of business attributes and travel patterns.

[0025] In this embodiment, edges are the connecting links between two nodes (initial aggregation service scope units) in the graph structure, carrying the spatial adjacency relationship between units. Semantic similarity is used as the edge weight; the larger the weight, the closer the spatial and business association between the two units. The graph structure is an undirected weighted graph constructed with initial aggregation service scope units as nodes, spatial adjacency relationships between units as the basis of association, and semantic similarity as edge weight. It is used to intuitively present the spatial distribution and business association characteristics of all initial units, providing a framework for the subsequent construction of the graph structure aggregation model.

[0026] S105: Construct a graph-structured aggregation model based on graph structure. The graph-structured aggregation model includes an objective function that aims to minimize the variance of statistical indicators among the initial aggregation service scope units, and constraints based on maintaining the integrity of spatial partitioning constraints.

[0027] In this embodiment, the construction is based on the generated undirected weighted graph structure. Based on spatial aggregation requirements, an objective function is defined, constraints are set, and a mathematical optimization model is built for iterative solution. This is the modeling step for achieving dynamic aggregation of spatial elements. The graph structure is an undirected weighted graph with initial aggregation service range units as nodes and spatial adjacency relationships and semantic similarity between units as edges. It serves as the basic carrier for constructing the graph structure aggregation model, providing node, edge, and relational data. The graph structure aggregation model is a mathematical optimization model built based on the graph structure. It is used to merge and reorganize the initial aggregation service range units, guides the aggregation direction through the objective function, limits the aggregation boundary through constraints, and finally outputs an aggregation result that meets the requirements.

[0028] Specifically, the graph-structured aggregation model adopts a multi-level spatial aggregation architecture, consisting of a spatial partitioning unit node layer, a spatial adjacency and semantic association edge layer, a statistical equilibrium target layer, a multi-constraint constraint layer, and an iterative optimization solution layer. The spatial partitioning unit node layer carries the initial aggregation service scope units and their statistical indicators. The spatial adjacency and semantic association edge layer represents the spatial adjacency relationship between units and the semantic similarity of supply and demand, land use function, and interaction flow. The statistical equilibrium target layer constructs an objective function with the goal of minimizing the variance of statistical indicators among units. The multi-constraint constraint layer accesses and maintains external constraints such as spatial partitioning constraints and statistical equilibrium constraints. The iterative optimization solution layer uses heuristic search algorithms or graph partitioning algorithms to merge and reorganize units. The various levels form a closed-loop optimization structure through data transfer and constraint propagation, jointly completing the dynamic aggregation of spatial elements for shared bicycle parking points.

[0029] In this embodiment, the objective function is a component of the graph-structured aggregation model and is a mathematical expression used to define the aggregation optimization objective. The core objective in this embodiment is to minimize the variance of statistical indicators among the initial aggregation service scope units, guiding the aggregation process towards statistical equilibrium. Minimization is achieved through iterative solutions, aiming to minimize the variance of statistical indicators among the initial aggregation service scope units, with the goal of minimizing the differences in statistical indicators (e.g., capacity-demand ratio) among the aggregated units, thus achieving supply-demand balance.

[0030] In this embodiment, statistical indicators are characteristic indicators of each initial aggregation service scope unit, with the core being the capacity-demand ratio. These also include the total number of parking spots, total capacity, and historical total demand, and are the data used to calculate variance and construct the objective function. Variance is a statistic used to represent the dispersion of a set of data (statistical indicators of each initial unit). The smaller the variance, the more balanced the statistical indicators of each unit. In this embodiment, it serves as the optimization object of the objective function. Constraints are rules used in the graph-structured aggregation model to limit the optimization scope and regulate aggregation behavior. In this embodiment, the constraint is to maintain the integrity of spatial partitioning constraints, ensuring that the aggregation process does not violate preset spatial boundary rules. Spatial partitioning constraints are user-defined spatial partitioning boundary rules (e.g., regular grids, administrative divisions, natural street boundaries), which are rigid / semi-rigid constraints that cannot be broken during the aggregation process. The integrity of spatial partitioning constraints means that during the aggregation process, the merging and reorganization of initial aggregation service scope units must not arbitrarily break the boundaries of spatial partitioning constraints. For example, they must not cross administrative division boundaries or disrupt the alignment of regular grids, ensuring that the spatial boundaries of the aggregation result meet preset management requirements.

[0031] S106: Use heuristic search algorithms or graph partitioning algorithms to iteratively solve the graph structure aggregation model and partition it. Under the constraint of spatial partitioning, merge or reorganize adjacent initial aggregation service range units to generate dynamic aggregation service range results that satisfy statistical equilibrium constraints.

[0032] In this embodiment, the heuristic search algorithm is an optimization algorithm based on empirical rules. This embodiment uses the simulated annealing algorithm to quickly search for the optimal or near-optimal unit partitioning scheme in large-scale graph structures, featuring fast iterative convergence and adaptability to spatial aggregation scenarios. The graph partitioning algorithm is a dedicated solution algorithm for partitioning and cutting graph structures. This embodiment uses the METIS algorithm to divide the nodes in the graph structure into several connected, balanced subgraphs that satisfy constraints.

[0033] Specifically, the heuristic search algorithm for dynamic aggregation of spatial elements adopts a multi-layer iterative optimization architecture, which consists of an initial solution generation module, a neighborhood search module, a target evaluation module, a constraint verification module, and an update convergence module. The initial solution generation module generates an initial partitioning scheme based on the initial aggregation service range units. The neighborhood search module performs neighborhood operations such as merging and splitting adjacent units. The target evaluation module calculates the fitness value based on the variance of the statistical indicators of each unit. The constraint verification module is used to determine whether the current partitioning satisfies the spatial partitioning constraints and statistical equilibrium constraints. The update convergence module retains the optimal solution based on the fitness and determines whether the termination condition has been reached. All modules work together to form a closed-loop iterative structure for balanced aggregation of shared bicycle service ranges.

[0034] Specifically, the graph partitioning algorithm for dynamic aggregation of spatial elements adopts a four-level structure: spatial topology graph construction layer, partition optimization layer, constraint verification layer, and result output layer. The spatial topology graph construction layer constructs an undirected weighted graph with the initial aggregation service range unit as the node and spatial adjacency and semantic similarity as the edge weight. The partition optimization layer clusters the nodes and partitions the regions based on the graph partitioning criteria. The constraint verification layer judges the compliance of the partitioning results with spatial partitioning constraints and statistical equilibrium constraints. The result output layer performs boundary normalization and region merging on the partitions that meet the constraints. Each level works in sequence to achieve a balanced partitioning of the service range for shared bicycle parking points.

[0035] In this embodiment, iterative solution partitioning is a process of gradually approaching the optimal solution by repeatedly adjusting, merging, and reorganizing units under the constraints of the objective function. Each iteration updates the unit combination and verifies the constraint satisfaction. Spatial partitioning constraints are user-preset rigid rules for spatial boundaries, including administrative divisions, natural blocks, and regular grid boundaries, which must be strictly adhered to during the iteration process to ensure the integrity of the boundaries is not compromised. Merging or reorganizing is the operation of combining adjacent initial aggregated service range units that satisfy adjacency relationships, have high semantic similarity, and are conducive to equilibrium indicators into a larger aggregated service unit; it is a method for implementing dynamic aggregation.

[0036] In this embodiment, adjacent initial aggregation service range units are node units in the graph structure that are directly connected by edges and have spatial adjacency relationships; these are the only objects allowed to be merged. The statistical equilibrium constraint is the requirement for minimizing the variance of the statistical indicators (capacity demand ratio) of each unit after aggregation and ensuring that the differences in indicators are within a reasonable range; this is the goal of optimizing the graph structure aggregation model. The dynamic aggregation service range result is the final service partition set generated after iterative solution. It possesses the characteristics of spatial regularity, boundary compliance, supply and demand balance, and real-time updates with data, providing a partitioning basis for the scheduling, management, and display of shared bicycles.

[0037] As can be seen from the above, this application obtains a dataset of spatial elements of shared bicycle parking points and multidimensional constraints, generates initial aggregated service range units based on spatial partitioning constraints, maps elements and calculates statistical indicators, constructs a graph structure with units as nodes and adjacency relationships and semantic similarity as edges, establishes an aggregation model with the goal of minimizing variance and the constraint of complete spatial constraints, and finally iteratively solves the merging and reorganizing of units through heuristic or graph partitioning algorithms to generate a dynamic aggregated service range that satisfies statistical equilibrium. This can achieve balanced division of service areas while ensuring compliance with spatial boundaries, improve the rationality, balance and practicality of shared bicycle service range division, and thus improve the efficiency of regional scheduling and management.

[0038] In one embodiment of this application, based on spatial partitioning constraints, the target research region is spatially divided to generate several initial aggregated service range units with basic attributes, including: Obtain the geographic boundary data corresponding to the spatial partitioning constraints, and use it as the spatial partitioning constraint boundary; If the spatial partitioning constraint is a regular grid, then a regular grid covering the target study area is generated, and the regular grid is overlaid with the spatial partitioning constraint boundary for analysis. The boundary of the regular grid is adjusted to align with the spatial partitioning constraint boundary, and each adjusted regular grid is used as a candidate initial unit. If the spatial division constraint is the boundary of an administrative division or a natural block, then each closed boundary is directly used as a candidate initial unit. The shared bicycle parking points that fall within each candidate initial unit are counted to form the initial aggregated service range units; For candidate initial cells that do not include any shared bicycle parking spots, they are merged or marked as empty cells based on the attributes of neighboring cells.

[0039] In this embodiment, spatial partitioning constraints are user-defined constraints used to limit the form and boundaries of regional spatial subdivision, including regular grid constraints, administrative division constraints, and natural block boundary constraints. Geographic boundary data are vector geographic data describing spatial extent, administrative boundaries, block outlines, and grid extents, consisting of coordinate strings forming closed boundaries, and serving as the basis for spatial subdivision. Spatial partitioning constraint boundaries are rigid or semi-rigid geographic boundaries determined by spatial partitioning constraints and cannot be arbitrarily broken, used to limit the generation range and shape of initial units.

[0040] In this embodiment, the regular mesh is a rectangular regular partitioning unit generated according to a preset size (e.g., 50m×50m, 100m×100m), used to uniformly divide the study area spatially. Overlay analysis is a spatial analysis operation that spatially overlays, clips, and matches the regular mesh layer with the spatial partitioning constraint boundary layer to achieve mesh boundary alignment. Overlay alignment is a process of matching, clipping, and correcting the regular mesh boundary with the spatial partitioning constraint boundary to ensure that the outer boundary of the mesh remains consistent with the constraint boundary. The candidate initial unit is a temporary spatial unit obtained after spatial partitioning that has not yet undergone parking point statistics and empty unit processing; it is the precursor to the initial aggregated service range unit.

[0041] In this embodiment, administrative boundaries are legally closed geographical boundaries formed by administrative levels such as streets, townships, communities, and districts, and are considered rigid spatial constraints. Natural street boundaries are street-style closed boundaries formed by natural geographical elements such as urban road networks, rivers, green spaces, and railways. Closed boundaries are spatial geometric boundaries where the coordinates of the beginning and end are connected, forming a closed area without intersections, and can be directly used as independent spatial units.

[0042] In this embodiment, the statistical inclusion is a process of calculating the candidate initial unit for each shared bicycle parking point through spatial inclusion judgment. A candidate initial unit that does not include any shared bicycle parking points and has no parking point elements included is called an empty unit. Neighboring unit attributes are the attributes of the initial aggregated service area units spatially adjacent to the empty unit, such as their number, land use type, and statistical indicators. Merging involves merging empty units without parking points into adjacent non-empty units, so that they no longer participate in subsequent calculations as independent units. Marking as an empty unit involves identifying candidate initial units without parking points, preventing them from participating in indicator calculations and graph structure construction. The initial aggregated service area unit is a formal basic spatial unit with basic attributes, formed after partitioning, parking point statistics, and empty unit processing.

[0043] As can be seen from the above, this embodiment generates candidate initial units by using regular grid alignment or directly using closed boundaries to divide spatial constraints, counts parking points within units, and merges or marks empty units. This can adapt to various spatial constraint scenarios such as administrative divisions, natural blocks, and regular grids, ensuring that the boundaries of the initial aggregation service range units are regular and highly consistent with the constraint boundaries. At the same time, invalid empty units are eliminated, which improves the simplification and accuracy of subsequent spatial mapping and index calculation, thereby enhancing the stability and reliability of the overall spatial aggregation process.

[0044] In one embodiment of this application, features in a spatial feature dataset are mapped to initial aggregation service range units, and statistical indicators for each initial aggregation service range unit are calculated, including: Based on spatial indexing or spatial connection operations, each shared bicycle parking point is mapped to the initial aggregated service range unit within the corresponding spatial partitioning constraint boundary; The total number of shared bicycle parking spots within each initial aggregation service unit is counted and used as the first statistical indicator. Obtain the capacity attribute of each shared bicycle parking point, and sum them up to get the total capacity of each initial aggregated service range unit, which is used as the second statistical indicator; Historical demand data for each shared bicycle parking spot is obtained and summed to obtain the total historical demand for each initial aggregated service area unit, which is used as the third statistical indicator; Based on the second and third statistical indicators, the capacity demand ratio of each initial aggregated service range unit is calculated as the final statistical indicator.

[0045] In this embodiment, the spatial index is a fast retrieval structure established to improve spatial matching efficiency, used to quickly locate the unit to which the parking point belongs. The spatial connection operation is a spatial analysis operation based on spatial location relationships, associating shared bicycle parking point elements with the initial aggregated service area unit. The spatial partitioning constraint boundary is a rigid / semi-rigid spatial boundary preset by the user, such as a grid, administrative division, or natural street. The initial aggregated service area unit is a basic spatial unit obtained through spatial partitioning, used to carry parking points and statistical indicators.

[0046] In this embodiment, statistical indicators are quantitative values ​​used to describe the supply and demand characteristics of a unit and provide a basis for aggregation equilibrium. The first statistical indicator is the total number of shared bicycle parking spots included within each initial aggregation service unit. The capacity attribute is the maximum number of shared bicycles that a parking spot can accommodate. The second statistical indicator is the total capacity obtained by summing the capacity attributes of all parking spots within the unit. Historical demand data refers to data representing demand intensity, such as the frequency of bicycle borrowing and returning and usage intensity, within a preset time window. The third statistical indicator is the historical total demand obtained by summing the historical demand data of all parking spots within the unit. The capacity-demand ratio is the ratio calculated from the total capacity and the historical total demand, serving as the final statistical indicator characterizing the degree of supply-demand matching within the unit. These final statistical indicators are used for subsequent graph construction, objective function optimization, and equilibrium judgment.

[0047] As can be seen from the above, this embodiment accurately maps parking points to the initial aggregation service range units through spatial indexing or spatial connection, sequentially counts the total number of parking points, the accumulated total capacity, and the historical total demand, and calculates the capacity-demand ratio based on capacity and demand as the final statistical indicator. This can comprehensively quantify the supply and demand characteristics of each unit, provide an objective and unified quantitative basis for graph construction and model optimization, improve the representativeness and rationality of statistical indicators, and thus enhance the scientific nature of the aggregation model solution and the usability of the final partitioning results.

[0048] In one embodiment of this application, the second and third statistical indicators, used to calculate the capacity demand ratio of each initial aggregated service range unit as the final statistical indicator, include: Obtain historical bike borrowing and returning data within a preset time window, and separately count the number of bikes borrowed and returned at each bike-sharing parking point; The historical demand weight value of the corresponding shared bicycle parking point is obtained by weighting and summing the number of bikes borrowed and returned at each parking point. The historical demand weights of all shared bicycle parking spots within the same initial aggregation service range unit are summed to obtain the historical total demand for the corresponding initial aggregation service range unit. Divide the total capacity of the corresponding initial aggregated service scope unit by the historical total demand to obtain the capacity demand ratio.

[0049] In this embodiment, the preset time window is a pre-defined time period for extracting historical bike-sharing data, such as the past 7 days, the past 30 days, or weekday / weekend periods. Historical bike-sharing data refers to the records of bikes being borrowed and returned at each shared bike parking spot within the preset time window. The borrowing count is the total number of times / number of bikes borrowed by users at a single shared bike parking spot within the preset time window. The returning count is the total number of times / number of bikes returned by users at a single shared bike parking spot within the preset time window. Weighted summation is a calculation method that weights the borrowing and returning counts separately according to preset weight coefficients before summing them.

[0050] In this embodiment, the historical demand weight value is a value obtained by weighted summing of borrowing and returning amounts, used to characterize the demand intensity of a single parking spot. The historical total demand is the cumulative result of the historical demand weight values ​​of all parking spots within the same initial aggregated service area unit, i.e., the third statistical indicator. The total capacity is the cumulative result of the capacity attributes of all parking spots within the same initial aggregated service area unit, i.e., the second statistical indicator.

[0051] As can be seen from the above, this embodiment obtains historical car rental and return data within a preset time window, calculates the demand weight by weighted summation of the rental and return amounts, accumulates the total historical demand of the unit, and uses the ratio of the total capacity to the total historical demand as the capacity demand ratio. This can more realistically represent the actual demand intensity of parking points and units, avoid the deviation caused by simple accumulation, improve the accuracy of historical demand assessment and capacity demand ratio calculation, and enhance the rationality of balance judgment and partition optimization in the aggregation process.

[0052] In one embodiment of this application, a graph structure is constructed with initial aggregation service range units as nodes and spatial adjacency relationships and semantic similarity between units as edges, including: Based on the geometric boundaries of the initial aggregated service range units, adjacent units with common edges or common vertices are identified, and spatial adjacency relationships are established. Calculate the semantic similarity of each pair of adjacent units, and use it as the weight of the corresponding edge; For spatially non-adjacent but semantically related units, if the spatial distance is less than a preset spatial threshold, virtual edges are established and assigned corresponding weights. Construct an undirected weighted graph structure based on nodes and edges.

[0053] In this embodiment, the geometric boundary is the spatial outline boundary of the initial aggregation service range unit, a closed geometric figure composed of a series of coordinate points, used to identify the spatial adjacency relationship between units. A common edge is a line segment that completely overlaps in the geometric boundaries of two initial aggregation service range units, representing the primary form of spatial adjacency between units. A common vertex is an endpoint that overlaps in the geometric boundaries of two initial aggregation service range units, representing a secondary form of spatial adjacency between units. Adjacent units are two initial aggregation service range units whose geometric boundaries share a common edge or a common vertex, forming a pair of adjacent units.

[0054] In this embodiment, an adjacent unit pair is a combination of two spatially adjacent initial aggregation service range units, serving as the basic unit for calculating semantic similarity and establishing spatial adjacency edges. The edge weight is a quantified value assigned to each edge in the graph structure, representing the degree of connection between two nodes (units); in this embodiment, it represents the semantic similarity of adjacent unit pairs. Strong semantic connection refers to two units that are closely related in terms of business attributes, such as frequent historical borrowing and returning interactions, highly matched supply and demand characteristics, and complementary land use functions, but are not spatially adjacent. The preset spatial threshold is a pre-set spatial distance threshold used to determine whether spatially non-adjacent units meet the conditions for establishing virtual edges. Virtual edges are specially established for units that are not spatially adjacent but have strong semantic connections and whose spatial distance is less than the preset spatial threshold. They are used to reflect the business connection between units and are distinct from spatial adjacency edges. The corresponding weight is a quantified weight assigned to the virtual edge, set according to the semantic connection strength of the two units (e.g., historical borrowing and returning interaction volume). The weight value is less than the average weight of spatially adjacent edges, reflecting the degree of connection. The undirected weighted graph structure is an undirected weighted graph constructed with the initial aggregated service scope unit as the node, spatial adjacency edges and virtual edges as the connection links, and semantic similarity (or association strength) as the edge weight. It is used to present the spatial and business associations of all units and provide a basic framework for the subsequent graph structure aggregation model construction.

[0055] The method for determining the preset spatial threshold is as follows: calculate the average equivalent diameter of each initial aggregated service range unit, combine it with the historical cycling trajectory to obtain the typical cycling radius of the area, take the larger value of the two and multiply it by the adaptive coefficient to obtain the basic spatial threshold; then introduce a hotspot correction factor according to the distribution of cycling hotspot areas, and finally obtain the adaptive preset spatial threshold, which is used to determine whether virtual edges are established for units that are not spatially adjacent but semantically strongly related.

[0056] As can be seen from the above, this embodiment establishes spatial adjacency edges by identifying common edges and common vertices of units, calculates semantic similarity as edge weights, and establishes virtual edges for units that are not spatially adjacent but have strong semantic connections and meet preset spatial thresholds, thus constructing an undirected weighted graph structure. This structure can simultaneously reflect spatial proximity relationships and business semantic connections, making the graph structure more in line with actual travel patterns, improving the graph structure's ability to express regional association features, and enhancing the accuracy and fit of the graph structure aggregation model division.

[0057] In one embodiment of this application, the semantic similarity of each pair of adjacent units is calculated as the weight of the corresponding edge, including: Calculate the absolute value of the difference between the capacity demand ratios of two adjacent units to obtain the supply-demand difference degree; Obtain the land use type of the area where two adjacent units are located; If the land use function types are the same, then assign the first score for functional similarity; If the land use types are complementary, a second score for functional similarity is assigned. Obtain the historical borrowing and returning traffic between two adjacent units, and then normalize it to obtain the interaction intensity value; The semantic similarity is obtained by weighting and summing the supply-demand difference, functional similarity, and interaction intensity values.

[0058] In this embodiment, adjacent unit pairs are two initial aggregated service range units that share a common edge or vertex in space and form an adjacency relationship. Semantic similarity is a 0-1 interval quantified value obtained by comprehensively considering supply-demand differences, land use functions, and travel interactions, used to represent the degree of business similarity between units and serving as the weight of the graph structure edges. The capacity-demand ratio is the ratio of the unit's total capacity to its historical total demand, characterizing the degree of supply-demand matching between units. The absolute value of the difference in capacity-demand ratios is the absolute value of the difference between the capacity-demand ratios of two adjacent units, used to represent the magnitude of the difference in supply and demand levels. The supply-demand difference degree is a quantified index obtained from the absolute value of the difference in capacity-demand ratios, used to characterize the degree of difference in supply and demand capabilities between two units.

[0059] In this embodiment, land use function type refers to the land use nature of the area where the unit is located, including residential, commercial, office, transportation hub, public service, green space, etc. Functional similarity is a quantitative score assigned based on the similarity or complementarity of the land use functions of two units, used to characterize the degree of functional association in the region. Similar functions mean that the land use function types of two adjacent units are completely identical. Complementary functions mean that the land use function types of two adjacent units are different but there is a strong correlation in travel, such as residential-commercial, residential-office. The first score is the functional similarity score assigned when the land use function types are the same. The second score is the functional similarity score assigned when the land use function types are complementary.

[0060] In this embodiment, historical bike-sharing traffic volume refers to the total number of shared bike interactions within a preset time window, where bikes are borrowed from one unit and returned to another. Normalization is a process that linearly maps historical bike-sharing traffic volume to the 0-1 range, eliminating the influence of dimensions. The interaction intensity value is a 0-1 value obtained after normalizing the historical bike-sharing traffic volume, representing the degree of closeness of cycling trips between two units. Weighted summation is performed by weighting the supply-demand difference, functional similarity, and interaction intensity value according to preset weight coefficients, and then summing them to obtain the final semantic similarity.

[0061] As can be seen from the above, this embodiment calculates the supply-demand difference, assigns functional similarity based on land use type, obtains interaction intensity value based on historical borrowing and returning flow, and obtains semantic similarity by weighted summation of the three. It can comprehensively quantify the similarity between units from multiple dimensions such as supply-demand balance, functional association, and travel interaction, improve the comprehensiveness and rationality of semantic similarity calculation, make the edge weight of graph structure more able to represent the real association strength, and improve the scientificity and practicality of aggregation partitioning.

[0062] In one embodiment of this application, a method for dynamic aggregation of spatial features based on multiple constraints further includes: Collect user interaction data and actual application effect data on the results of dynamic aggregation services; The actual application effect data is compared with the target value of statistical equilibrium constraint to obtain the aggregation effect deviation rate; Based on the deviation rate between interactive operation data and aggregation effect, the parameters of multidimensional constraints and / or the weight coefficients of the objective function are adjusted to obtain the updated graph structure aggregation model. Dynamic aggregation of shared bicycle parking point service areas is performed based on the updated graph structure aggregation model.

[0063] In this embodiment, interactive operation data refers to the recorded data generated when users perform operations such as viewing, adjusting, dragging, selecting, zooming in, zooming out, and modifying parameters on the results of the dynamically aggregated service scope in the system interface. This data reflects the user's usage intent and optimization preferences. Actual application effect data refers to the real effect data generated during the actual scheduling, operation, and display of the dynamically aggregated service scope, including regional balance, scheduling costs, parking spot utilization, and user satisfaction. The target value of the statistical balance constraint is a pre-set ideal target value used to measure the degree of balance of statistical indicators of each unit after aggregation, such as the target variance, target mean, and target difference threshold of the capacity demand ratio. The aggregation effect deviation rate is a quantitative deviation value calculated by comparing the actual application effect data with the statistical balance constraint target value, used to characterize the degree of deviation between the current aggregation result and the ideal target.

[0064] In this embodiment, the parameters of the multidimensional constraints include adjustable parameters such as boundary parameters of spatial partitioning constraints, mesh size, equilibrium constraint threshold, and aggregation granularity. The weight coefficients of the objective function are the coefficients used to control the proportion of different optimization objectives such as variance minimization, spatial regularity, and equilibrium. Adjustment is the process of correcting and updating the constraint parameters and weight coefficients based on the interaction intent and effect deviation, making the graph structure aggregation model more closely match actual needs. The updated graph structure aggregation model is a new, optimized version of the aggregation model formed after parameter and weight adjustments. Dynamic aggregation is based on the updated graph structure aggregation model, re-executing unit merging, reorganization, and solving to generate service scope results that better meet actual needs.

[0065] As can be seen from the above, this embodiment collects user interaction data and actual application effect data, calculates the aggregation effect deviation rate, and adaptively adjusts the multidimensional constraint parameters and objective function weights accordingly. It updates the graph structure aggregation model and re-executes dynamic aggregation, which can achieve closed-loop optimization and self-iteration of the model. This makes the aggregation results continuously fit the actual operation needs, improves the adaptive ability and actual application effect of the dynamic aggregation method, and enhances the intelligence level of shared bicycle scheduling and management.

[0066] In one embodiment of this application, a method for dynamic aggregation of spatial features based on multiple constraints further includes: Based on real-time supply and demand data or map view zoom level, adjust the aggregation granularity, trigger the re-execution of the iterative solution partitioning step, and obtain the updated dynamic aggregation service range result; When a change in the map view zoom level is detected, the current view zoom level is obtained and matched with a preset granularity level mapping table to determine the target aggregation granularity level. Based on the target aggregation granularity level, adjust the threshold range of statistical equilibrium constraints and the weight coefficients for minimizing variance in the objective function; When real-time supply and demand data updates are detected, the statistical indicators of each unit are recalculated based on the latest parking point borrowing and returning data, the node weights and edge weights are updated, and the graph structure aggregation model is re-solved. When it is detected that a user has selected a specific area for focused analysis, the priority of the statistical equilibrium constraint within the selected area is temporarily increased, while the strictness of the spatial division constraint outside the selected area is reduced.

[0067] In this embodiment, real-time supply and demand data refers to operational data such as the number of shared bicycle parking spots borrowed and returned, available capacity, and demand intensity acquired at the current or near real-time moment, used to represent the immediate supply and demand status. Map view zoom level refers to the zoom level when displaying the electronic map; different zoom levels correspond to different spatial display ranges and levels of detail. Aggregation granularity refers to the coarseness of the dynamically aggregated service range; larger granularity indicates a coarser partition and a larger range, while smaller granularity indicates a finer partition and a smaller range. Iterative solution partitioning is an iterative optimization process based on a graph-structured aggregation model, merging and reorganizing the initial aggregated service range units. The updated dynamic aggregated service range result is a service range partitioning result adapted to the current scenario, obtained by resolving based on aggregation granularity, real-time data, or user operations.

[0068] In this embodiment, the granularity level mapping table is a pre-established correspondence table between map zoom levels and aggregation granularity levels, used to automatically match the aggregation granularity based on view zoom. The target aggregation granularity level is the aggregation granularity level that matches the current map zoom level, generally divided into three levels: macro, meso, and micro. The threshold range of the statistical equilibrium constraint is the upper and lower limits used to control the magnitude of differences in statistical indicators among the aggregated units. The weighting coefficient for minimizing variance is a coefficient in the objective function used to control the proportion of the statistical equilibrium objective.

[0069] In this embodiment, node weights and edge weights are the statistical index weights of nodes and the semantic similarity weights of edges in the graph structure, respectively. Re-solving involves re-executing the iterative optimization process of the graph structure aggregation model based on the updated indices, parameters, and constraints. User-selected specific areas refer to the key analysis areas specified by the user on the map interface through box selection, circle selection, or other operations. Focused analysis is an analysis mode that uses the user-selected area as the core, strengthening local equilibrium and weakening the strictness of external constraints. The priority of statistical equilibrium constraints refers to the degree of priority of statistical index equilibrium requirements within a local area. The strictness of spatial division constraints refers to the constraint strength on the insurmountability of spatial boundaries such as administrative divisions, grids, and street blocks.

[0070] As can be seen from the above, this embodiment dynamically adjusts the aggregation granularity based on real-time supply and demand data or map zoom level, matches the granularity level and adjusts the constraint threshold and target weight accordingly, recalculates the indicators and resolves them when supply and demand are updated, and increases the priority of local equilibrium and reduces the strictness of external constraints when users select areas. This enables adaptive dynamic aggregation under multiple scales and scenarios, improves the service range's responsiveness to view changes, real-time data and user operations, and thus enhances the system's user experience and analysis efficiency.

[0071] In one embodiment of this application, adjusting the threshold range of statistical equilibrium constraints and the weight coefficients for minimizing variance in the objective function based on the target aggregation granularity level includes: If the target aggregation granularity level is macroscopic, then the upper limit of the threshold of statistical equilibrium constraint is increased based on the first adjustment step size, while the weight of minimizing variance in the objective function is decreased based on the second adjustment step size. If the target aggregation granularity level is meso-level, then the default parameters of statistical equilibrium constraints and objective function weights are maintained. If the target aggregation granularity level is micro-level, then the threshold lower limit of the statistical equilibrium constraint is tightened based on the third adjustment step size, while the weight of minimizing variance in the objective function is increased based on the fourth adjustment step size.

[0072] In this embodiment, the upper threshold is the maximum allowable difference in statistical indicators under the statistical equilibrium constraint; a larger upper threshold indicates greater allowable imbalance. The lower threshold is the minimum allowable difference in statistical indicators under the statistical equilibrium constraint; a smaller lower threshold indicates stricter requirements. The objective function graph is the optimization function in the graph-structured aggregation model that focuses on minimizing the variance of statistical indicators in each unit. The weight coefficients for variance minimization are coefficients used within the objective function to control the proportion of statistical equilibrium objectives; a larger weight indicates a greater pursuit of equilibrium. The macro level is a large-scale, overview-style aggregation granularity with few partitions and a large scope, emphasizing spatial integrity. The meso level is the conventional default aggregation granularity, balancing spatial constraints and statistical equilibrium. The micro level is a small-scale, refined aggregation granularity with many partitions and a small scope, emphasizing high equilibrium.

[0073] In this embodiment, the first adjustment step size is at the macro level and is specifically used to increase the preset adjustment range of the upper limit of the statistical equilibrium constraint threshold. The second adjustment step size is at the macro level and is specifically used to decrease the preset adjustment range of the variance minimization weight. The third adjustment step size is at the micro level and is specifically used to tighten the preset adjustment range of the lower limit of the statistical equilibrium constraint threshold. The fourth adjustment step size is at the micro level and is specifically used to increase the preset adjustment range of the variance minimization weight.

[0074] The first adjustment step size is calculated based on the tolerance accuracy of the regional overview in the macro view, the default threshold range of the statistical equilibrium constraint, the map zoom level, and the view area ratio. Using the upper limit of the default threshold of the statistical equilibrium constraint as a benchmark, the threshold increment that can ensure the clarity of the overall trend of the region without excessively disrupting the equilibrium at the macro level is determined through pre-experimental statistical analysis of multiple regions at different scales. This increment is normalized and used as the basic step size. It is then corrected by the total area and number of units of the target study area to finally obtain the first adjustment step size used to increase the upper limit of the statistical equilibrium constraint threshold at the macro level.

[0075] The second adjustment step size is calculated based on the default value range of the variance minimization weight coefficient in the objective function, the need to weaken the balance constraint of macro-level aggregation, and the convergence stability and regional display effect of the algorithm in multiple pre-experiments. Taking the default value of the variance minimization weight coefficient as the benchmark, 5%-15% of its value range is selected as the single adjustment range. This allows for an appropriate reduction in the balance constraint weight at the macro level, strengthening spatial integrity and view simplicity, while avoiding excessively rapid weight reduction that could lead to imbalance in the aggregation results or algorithm oscillation. The final result is a second adjustment step size that adapts to the macro view granularity and is used to reduce the variance minimization weight.

[0076] The third adjustment step size is calculated based on the default threshold range of statistical equilibrium constraints, the accuracy requirements of supply and demand equilibrium at the micro level, and the number of units and the fluctuation range of demand under the micro view. Taking the lower limit of the default threshold of statistical equilibrium constraints as a benchmark, the threshold contraction range that meets the needs of fine management is determined through multiple sets of micro-scale pre-experiments. This range is set to 5%-15% of the default threshold range, so that the equilibrium constraints can be significantly tightened at the micro level and the rationality of the service scope can be improved. At the same time, the constraint effect will not be insignificant due to the small range. Finally, the third adjustment step size for tightening the lower limit of the statistical equilibrium constraint threshold at the micro level is obtained.

[0077] The fourth adjustment step size is calculated based on the range of values ​​for the variance minimization weight coefficient in the objective function, the strict requirements of the micro-level for statistical equilibrium, and the impact of weight increase on aggregation accuracy and convergence stability in multiple pre-experiments. Using the default value of the variance minimization weight coefficient as a benchmark, 5%-15% of its effective value range is selected as the single adjustment range. This allows for a significant strengthening of statistical equilibrium constraints and an improvement in the supply and demand balance among service range units at the micro-level, while avoiding excessive adjustment that could cause algorithm oscillation or overly fragmented partitioning results. Ultimately, a fourth adjustment step size is obtained that is adapted to the micro-view granularity and used to improve the variance minimization weight coefficient.

[0078] In this embodiment, the default parameters are standard parameters pre-set by the system and applicable to the meso-level, without any additional adjustments. Increasing the upper limit of the threshold relaxes the equilibrium requirements, allowing greater differences in statistical indicators between units. Tightening the lower limit of the threshold increases the equilibrium requirements, forcing smaller differences in statistical indicators between units.

[0079] As can be seen from the above, this embodiment adjusts the statistical equilibrium constraints by increasing the threshold, maintaining the default, and tightening the threshold at the three aggregation granularity levels of macro, meso, and micro, respectively, and adjusts the variance minimization weight accordingly. This can reasonably balance spatial regularity and statistical equilibrium at different display scales, ensuring an overall overview at the macro level and strengthening local equilibrium at the micro level, thereby improving the rationality and applicability of multi-scale aggregation results and enhancing the effectiveness of map visualization and regional analysis.

[0080] In one embodiment of this application, before spatially partitioning the target research area based on spatial partitioning constraints to generate several initial aggregated service range units with basic attributes, the method further includes: Acquire historical cycling trajectory data for the target study area and identify naturally formed cycling hotspots; By overlaying the boundaries of cycling hotspots with spatial division constraints, inconsistent areas can be identified. The boundaries of the initial aggregated service range units within the inconsistent region are blurred to generate flexible boundary units with membership weights. The shared bicycle parking points within the flexible boundary unit participate in the calculation of statistical indicators of multiple adjacent units according to the preset membership weight.

[0081] In this embodiment, historical cycling trajectory data refers to the spatiotemporal data of historical location points, travel paths, start and end points generated by users riding shared bicycles within the target study area. Cycling hotspots are high-demand, contiguous areas naturally formed by concentrated cycling demand, obtained through density analysis of historical cycling trajectories. Overlay analysis involves spatially overlaying, comparing, and matching the boundaries of cycling hotspots with spatial division constraint boundaries to determine their consistency. Inconsistent areas are those where the boundaries of cycling hotspots do not coincide with, match, or exhibit misalignment or separation from spatial division constraint boundaries (grids, administrative districts, blocks). Fuzziness processing softens and transitions rigid boundaries that do not conform to actual travel patterns, replacing rigid hard boundaries. Flexible boundary units are special spatial units with unfixed boundaries, a certain transition range, and can be assigned to multiple adjacent units based on membership degree. Membership degree weights are weight values ​​between 0 and 1, indicating the proportion of parking points within a flexible boundary unit that are assigned to a particular adjacent unit. The statistical indicators involved in the calculation of multiple adjacent units include parking capacity and demand within the flexible boundary unit. These are split and included in multiple adjacent units according to membership weights, and are no longer uniquely attributed to a single unit.

[0082] As can be seen from the above, this embodiment identifies cycling hotspot areas before spatial partitioning, overlays and analyzes areas inconsistent with the constraint boundaries, and then fuzzifies these areas to generate elastic boundary units with membership degrees. This allows parking points within a unit to participate in the calculation of multiple adjacent units according to their membership degrees. This approach can mitigate the mismatch between rigid boundaries and actual travel patterns, improve the fit between the initial unit partitioning and natural cycling needs, reduce statistical bias caused by boundary fragmentation, and enhance the rationality and accuracy of the aggregation results.

[0083] In one embodiment of this application, when it is detected that a user has selected a specific area for focus analysis, the method further includes: Obtain historical origin and destination data of cycling within a specific area selected by the user, and construct an OD matrix for travel demand within the area; Based on the OD matrix, identify the main flow direction and volume of travel demand within a specific area selected by the user; Based on the main flow direction, a secondary aggregation with flow direction constraints is performed on the initial aggregation service range units within the user-selected specific area, so that the boundary of the aggregation result is aligned with the vertical direction of the main travel flow direction. The aggregated results after flow alignment are highlighted to help users analyze travel patterns within the region.

[0084] In this embodiment, the user-selected specific area refers to a local geographical region that the user specifies on the map interface through methods such as box selection, hand drawing, or circle selection, indicating a region requiring focused analysis. Focused analysis uses the user-selected area as the core, employing a refined analysis mode that prioritizes local constraints, enhances balance, and improves the matching degree of travel patterns. Historical ride start-end point data refers to the start and end locations of shared bicycle ride orders within a preset time window, abbreviated as OD data. Travel demand is represented by the OD matrix, a two-dimensional matrix with each initial aggregated service area unit within the region as rows and columns, and the travel volume between units as matrix elements, used to quantify travel interaction relationships within the region.

[0085] In this embodiment, the origin-destination matrix (OD matrix) describes the distribution of travel demand sources and destinations between spatial units. The primary flow direction is extracted from the OD matrix, representing the travel direction with the largest traffic volume and the most concentrated demand, indicating the dominant travel corridor within the selected area. Traffic volume is the total number of cycling orders occurring between two units per unit time, used to characterize travel intensity. The secondary aggregation of flow direction constraints is based on the initial aggregation, merging and reorganizing the initial units according to the primary travel flow direction, ensuring that the partition boundaries conform to travel patterns. The vertical direction of the primary travel flow direction is the direction perpendicular to the dominant travel corridor direction, used to align the aggregation boundaries, ensuring that the partition cutting direction matches the travel direction.

[0086] In this embodiment, aligning the aggregation result boundary perpendicular to the flow direction ensures that the boundary line of the final aggregation service area is perpendicular to the main travel direction, reducing cross-zone travel and improving the rationality of zoning. Highlighting rendering involves enhancing the visualization of the aligned aggregation boundary, zones, and flow corridors on the map interface through techniques such as highlighting, bolding, and outlining. Travel patterns represent the spatial distribution of cycling demand within the selected area, major corridors, and the intensity and direction of traffic flow.

[0087] As can be seen from the above, this embodiment constructs a travel OD matrix when the user selects a region for focused analysis, identifies the main flow direction and volume, and performs secondary aggregation to align the boundaries with the flow direction vertically. Then, it highlights and renders the results, which enables the aggregated partitions to conform to the real travel corridors, reduces the proportion of cross-regional travel, improves the pertinence and practicality of local area analysis, enhances the user's understanding of regional travel patterns and decision support capabilities, and thus improves the usability of the focused analysis function.

[0088] In one embodiment of this application, historical cycling trajectory data of the target study area is obtained, and naturally formed cycling hotspots are identified, including: Kernel density estimation is performed on historical cycling trajectory data to generate a cycling density distribution raster for the target study area; Extract grid cells with density values ​​greater than a preset density threshold and mark them as connected regions to obtain the initial hotspot regions; Morphological closing operations are performed on the initial hotspot region to fill the internal voids and smooth the boundaries, generating the final cycling hotspot region.

[0089] In this embodiment, historical cycling trajectory data refers to the spatiotemporal data of shared bicycle users within the target study area, including their riding trajectories, location points, and routes, generated within a preset time window. Cycling hotspots are continuous high-demand areas naturally formed by spatial analysis of historical cycling trajectories, representing highly concentrated cycling demand. Kernel density estimation is a spatial analysis method that calculates the density of spatial point data to generate a continuous, smooth density distribution surface, used to identify areas of concentrated demand. The cycling density distribution grid is raster data generated through kernel density estimation; each raster cell has a cycling density value, with higher values ​​indicating denser cycling.

[0090] In this embodiment, a grid cell is the smallest spatial unit of grid data, and each cell corresponds to a cycling density value. The preset density threshold is a pre-defined density threshold; only areas with a density greater than the preset density threshold are identified as cycling hotspots. Initial hotspot areas are the original hotspot areas obtained by connecting region labeling, where the density is greater than the preset density threshold, and they contain internal voids and rough boundaries. Connecting region labeling is the process of clustering spatially adjacent grid cells with densities all greater than the preset density threshold to form continuous regions.

[0091] Specifically, the method for determining the preset density threshold is as follows: First, all density values ​​of the cycling density distribution grid are statistically analyzed to obtain the density mean and density standard deviation. Then, the statistical threshold is obtained based on the adaptive multiple coefficient. At the same time, the 95th percentile threshold of all density values ​​is calculated. The larger value between the statistical threshold and the percentile threshold is used as the final preset density threshold to accurately extract cycling hotspot areas.

[0092] In this embodiment, morphological closing is a morphological operation in image processing, involving dilation followed by erosion, used to fill small holes and smooth the outer boundaries of regions. Filling internal holes eliminates small non-hotspot holes within the initial hotspot region, making the region more continuous and complete. Smoothing boundaries regularizes the jagged and fragmented boundaries of the hotspot region, making the contour more reasonable. The final cycling hotspot region is a standard hotspot region obtained after density calculation, threshold extraction, connected component labeling, and morphological processing, used for subsequent elastic boundary construction.

[0093] As can be seen from the above, this embodiment generates a cycling density raster by estimating the kernel density of historical cycling tracks, extracts high-density regions and marks connected components, and then generates cycling hotspot regions by filling holes and smoothing boundaries through morphological closing operations. This can accurately and robustly identify naturally formed high-demand regions, improve the continuity and regularity of hotspot region extraction, provide a reliable basis for subsequent flexible boundary construction, and thus enhance the overall method's adaptability to the real demand spatial distribution.

[0094] In one embodiment of this application, shared bicycle parking points within a flexible boundary unit participate in the calculation of statistical indicators for multiple adjacent units according to preset membership weights, including: For each shared bicycle parking spot within the elastic boundary unit, obtain the corresponding shared bicycle parking spot's capacity attribute and historical demand data; The weighted contribution value is obtained by multiplying the capacity and historical demand data of the corresponding shared bicycle parking point by the membership weight of the corresponding elastic boundary unit to each adjacent unit. The weighted contribution values ​​are added to the statistical indicators of the corresponding adjacent units; the original elastic boundary units are no longer used as independent statistical units in subsequent calculations.

[0095] In this embodiment, the flexible boundary unit is a transitional spatial unit within an area where cycling hotspots and spatial division constraints are inconsistent, after fuzzification and boundary flexibility. Shared bicycle parking spots are the shared bicycle parking locations included within the flexible boundary unit, representing the source of capacity and demand contribution. The preset membership weights are pre-defined weights between 0 and 1, characterizing the degree to which the flexible boundary unit belongs to its neighboring units; the sum of all weights is 1. The calculation of statistical indicators involving multiple neighboring units means that parking spots within the flexible boundary unit do not uniquely belong to any one unit, but are contributed to multiple neighboring units according to their weights.

[0096] In this embodiment, the capacity attribute is the maximum number of shared bicycles that a parking spot can accommodate. Historical demand data is the total historical borrowing and returning demand of the parking spot within a preset time window. The weighted contribution value is the split value obtained by multiplying the parking spot capacity and historical demand data by the membership weight. The contribution value accumulation is the accumulation of the weighted capacity and demand into the statistical indicators of the corresponding adjacent units. Adjacent units are the initial aggregation service range units that are spatially adjacent to the elastic boundary units. Statistical indicators include indicators used in the aggregation model, such as total unit capacity, total historical demand, and capacity-to-demand ratio. The elastic boundary unit itself is no longer considered an independent statistical unit because it does not participate in subsequent indicator calculations, graph construction, and model solving.

[0097] As can be seen from the above, this embodiment allocates the capacity and historical demand of parking points within the flexible boundary unit to adjacent units in a weighted manner according to the membership degree weight, and accumulates the contribution value to the corresponding unit's statistical index. At the same time, the flexible unit no longer participates in the calculation independently, which can realize the flexible attribution of the transition area, avoid the statistical distortion caused by hard boundaries, improve the authenticity and rationality of the calculation of the statistical index of each unit, and thus improve the reliability of the graph structure and aggregation model.

[0098] In one embodiment of this application, after generating the dynamic aggregation service range that satisfies the statistical equilibrium constraint, the method further includes: Boundary smoothing is performed on the dynamic aggregation service range results, including elastic boundary units; Identify virtual child nodes in the dynamic aggregation service range results that originate from the same elastic boundary unit but are divided into different aggregation ranges; Based on the membership weights and spatial distribution of virtual child nodes, a weighted average method is used to determine the final aggregation boundary line. The smoothed aggregation boundary line is output as the dynamic service range of shared bicycle parking points.

[0099] In this embodiment, the dynamic aggregation service range is the result of iteratively solving the graph structure aggregation model, satisfying both statistical equilibrium constraints and spatial partitioning constraints. Boundary smoothing is a spatial processing operation that regularizes and rounds the jagged, angular, and fragmented boundaries appearing in the dynamic aggregation service range. Elastic boundary units are transitional spatial units with membership weights, formed after fuzzification within areas where cycling hotspots and spatial partitioning constraints are inconsistent. Virtual child nodes are virtual affiliation units formed when the same elastic boundary unit is partitioned into different aggregation ranges according to its membership weights during the aggregation process.

[0100] In this embodiment, regions belonging to the same elastic boundary unit but assigned to different aggregation ranges are areas that originally belonged to the same elastic boundary unit but are assigned to different final service partitions after aggregation. The membership weight is a pre-defined value between 0 and 1, representing the degree to which a virtual child node belongs to a certain aggregation range. Spatial distribution of virtual child nodes refers to their geographical location, range, and adjacency relationships. The weighted average method uses the membership weight as a weighting coefficient to calculate a weighted average of spatial location and boundary coordinates to determine a continuous and reasonable boundary line. The final aggregation boundary line is a continuous and smooth closed boundary of the service range formed after weighted averaging and smoothing. The output is the final determined dynamic service range in the form of vector boundaries, layers, coordinate strings, etc., for map display and scheduling management. The dynamic service range is the final publicly released, directly usable, balanced, and well-defined shared bicycle parking service partition.

[0101] As can be seen from the above, this embodiment can eliminate jagged and fragmented boundaries by smoothing the boundaries of the dynamic aggregation results, identifying virtual child nodes formed by splitting the same elastic boundary, determining the final boundary based on the membership weight and spatial distribution weighted average, and outputting the smoothed service range. This makes the final partition outline continuous and smooth, improves the aesthetics and practicality of the dynamic aggregation service range, and facilitates map display and actual operation.

[0102] Corresponding to the spatial feature dynamic aggregation method based on multiple constraints in the above embodiment, Figure 2 This is a structural block diagram of a spatial feature dynamic aggregation system based on multiple constraints, provided as an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The spatial feature dynamic aggregation system 20 based on multiple constraints includes: a data acquisition module 21, a spatial subdivision module 22, an index calculation module 23, a graphics construction module 24, a model construction module 25, and an optimization solution module 26.

[0103] Among them, the data acquisition module 21 is used to acquire the spatial element dataset of shared bicycle parking points and user-defined multidimensional constraints; the multidimensional constraints include at least spatial partitioning constraints and statistical equilibrium constraints; Spatial partitioning module 22 is used to spatially partition the target study area based on spatial partitioning constraints, and generate several initial aggregation service range units with basic attributes. The indicator calculation module 23 is used to map the features in the spatial feature dataset to the initial aggregation service range units and calculate the statistical indicators of each initial aggregation service range unit. The graph construction module 24 is used to construct a graph structure with the initial aggregation service range unit as the node and the spatial adjacency relationship and semantic similarity between the units as the edge; The model building module 25 is used to build a graph structure aggregation model based on the graph structure. The graph structure aggregation model includes an objective function that aims to minimize the variance of statistical indicators among the initial aggregation service scope units, and constraints based on maintaining the integrity of spatial partitioning constraints. The optimization solution module 26 is used to iteratively solve the graph structure aggregation model using a heuristic search algorithm or graph partitioning algorithm. Under the constraint of spatial partitioning, it merges or reorganizes adjacent initial aggregation service range units to generate dynamic aggregation service range results that satisfy statistical equilibrium constraints.

[0104] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the data acquisition module 21, spatial partitioning module 22, index calculation module 23, graph construction module 24, model construction module 25, and optimization solution module 26 are shown.

[0105] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0106] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0107] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0108] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the spatial element dynamic aggregation method based on multiple constraints provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0109] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0110] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0111] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0112] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0113] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0114] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0115] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0116] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamic aggregation of spatial elements based on multiple constraints, characterized in that, include: Obtain the spatial element dataset of shared bicycle parking locations and user-defined multidimensional constraints; The multidimensional constraints include at least spatial partitioning constraints and statistical equilibrium constraints; Based on the aforementioned spatial partitioning constraints, the target research area is spatially divided to generate several initial aggregated service range units with basic attributes. The features in the spatial feature dataset are mapped to the initial aggregation service range units, and the statistical indicators of each initial aggregation service range unit are calculated. Construct a graph structure with the initial aggregation service range units as nodes and the spatial adjacency and semantic similarity between units as edges; Based on the graph structure, a graph structure aggregation model is constructed. The graph structure aggregation model includes an objective function that aims to minimize the variance of statistical indicators among the initial aggregation service scope units, and constraints based on maintaining the integrity of the spatial partitioning constraints. The graph structure aggregation model is iteratively solved and partitioned using a heuristic search algorithm or a graph partitioning algorithm. Under the spatial partitioning constraints, adjacent initial aggregation service range units are merged or reorganized to generate a dynamic aggregation service range result that satisfies the statistical equilibrium constraints.

2. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 1, characterized in that, Based on the spatial partitioning constraints, the target study area is spatially divided to generate several initial aggregated service range units with basic attributes, including: Obtain the geographic boundary data corresponding to the spatial partitioning constraints, and use it as the spatial partitioning constraint boundary; If the spatial partitioning constraint is a regular grid, then a regular grid covering the target study area is generated, and the regular grid is overlaid with the spatial partitioning constraint boundary for analysis. The boundary of the regular grid is adjusted to align with the spatial partitioning constraint boundary, and each adjusted regular grid is used as a candidate initial unit. If the spatial division constraint is an administrative division or a natural block boundary, then each closed boundary is directly used as a candidate initial unit; The shared bicycle parking points that fall into each candidate initial unit are counted to form an initial aggregated service range unit; For candidate initial units that do not include any of the aforementioned shared bicycle parking spots, they are merged or marked as empty units based on neighboring unit attributes.

3. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 1, characterized in that, The step of mapping features in the spatial feature dataset to the initial aggregation service range units and calculating statistical indicators for each initial aggregation service range unit includes: Based on spatial indexing or spatial connection operations, each shared bicycle parking point is mapped to the initial aggregated service range unit within the corresponding spatial partitioning constraint boundary; The total number of shared bicycle parking spots within each initial aggregation service range unit is counted and used as the first statistical indicator. Obtain the capacity attribute of each shared bicycle parking point, and sum them up to obtain the total capacity of each initial aggregated service range unit, which is used as the second statistical indicator; Historical demand data for each shared bicycle parking spot is obtained and summed to obtain the total historical demand for each initial aggregated service range unit, which is used as a third statistical indicator. Based on the second and third statistical indicators, the capacity demand ratio of each initial aggregation service range unit is calculated as the final statistical indicator.

4. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 3, characterized in that, Based on the second and third statistical indicators, the capacity demand ratio of each initial aggregated service range unit is calculated as the final statistical indicator, including: Obtain historical bike borrowing and returning data within a preset time window, and count the number of bikes borrowed and returned at each of the shared bike parking points. The borrowing and returning volumes of each shared bicycle parking spot are weighted and summed to obtain the historical demand weight value of the corresponding shared bicycle parking spot. The historical demand weight values ​​of all shared bicycle parking points within the same initial aggregation service range unit are summed to obtain the historical total demand for the corresponding initial aggregation service range unit. The capacity demand ratio is obtained by dividing the total capacity of the corresponding initial aggregated service scope unit by the historical total demand.

5. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 1, characterized in that, The construction of the graph structure, with the initial aggregation service range unit as nodes and the spatial adjacency relationship and semantic similarity between units as edges, includes: Based on the geometric boundaries of the initial aggregation service range unit, adjacent units with common edges or common vertices are identified, and spatial adjacency relationships are established. Calculate the semantic similarity of each pair of adjacent units, and use it as the weight of the corresponding edge; For spatially non-adjacent but semantically related units, if the spatial distance is less than a preset spatial threshold, virtual edges are established and assigned corresponding weights. Based on the nodes and edges, an undirected weighted graph structure is constructed.

6. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 5, characterized in that, The calculation of the semantic similarity of each adjacent unit pair, as the weight of the corresponding edge, includes: Calculate the absolute value of the difference between the capacity demand ratios of two adjacent units to obtain the supply-demand difference degree; Obtain the land use function type of the area where the two adjacent units are located; If the land use functions are the same, then a first score for functional similarity is assigned. If the land use functions are complementary, a second score for functional similarity is assigned. The historical vehicle borrowing and returning traffic between the two adjacent units is obtained, and the interaction intensity value is obtained after normalization. The semantic similarity is obtained by weighted summation of the supply-demand difference, functional similarity, and interaction intensity values.

7. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 1, characterized in that, Also includes: Collect user interaction data and actual application effect data on the results of the dynamic aggregation service. The actual application effect data is compared with the target value of the statistical equilibrium constraint to obtain the aggregation effect deviation rate; Based on the deviation rate between the interactive operation data and the aggregation effect, the parameters of the multidimensional constraints and / or the weight coefficients of the objective function are adjusted to obtain an updated graph structure aggregation model. Dynamic aggregation of shared bicycle parking point service areas is performed based on the updated graph structure aggregation model.

8. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 1, characterized in that, Also includes: Based on real-time supply and demand data or map view zoom level, adjust the aggregation granularity, trigger the re-execution of the iterative solution partitioning step, and obtain the updated dynamic aggregation service range result; When a change in the map view zoom level is detected, the current view zoom level is obtained and matched with a preset granularity level mapping table to determine the target aggregation granularity level. Based on the target aggregation granularity level, adjust the threshold range of the statistical equilibrium constraint and the weight coefficient for minimizing the variance of the objective function; When the real-time supply and demand data update is detected, the statistical indicators of each unit are recalculated based on the latest parking point borrowing and returning data, the node weights and edge weights are updated, and the graph structure aggregation model is re-solved. When it is detected that a user has selected a specific area for focused analysis, the priority of the statistical equilibrium constraint within the selected specific area is temporarily increased, and the strictness of the spatial division constraint outside the selected specific area is reduced.

9. The method for dynamic aggregation of spatial elements based on multiple constraints according to claim 8, characterized in that, The step of adjusting the threshold range of the statistical equilibrium constraint and the weight coefficients for minimizing the variance of the objective function based on the target aggregation granularity level includes: If the target aggregation granularity level is macroscopic, then the upper limit of the threshold of the statistical equilibrium constraint is increased based on the first adjustment step size, and the weight of minimizing variance in the objective function is decreased based on the second adjustment step size. If the target aggregation granularity level is meso-level, then the default parameters of the statistical equilibrium constraint and the objective function weights are maintained. If the target aggregation granularity level is micro-level, then the threshold lower limit of the statistical equilibrium constraint is tightened based on the third adjustment step size, and the weight of minimizing variance in the objective function is increased based on the fourth adjustment step size.

10. A spatial element dynamic aggregation system based on multiple constraints, characterized in that, include: The data acquisition module is used to acquire the spatial element dataset of shared bicycle parking points and user-defined multidimensional constraints. The multidimensional constraints include at least spatial partitioning constraints and statistical equilibrium constraints; The spatial partitioning module is used to spatially partition the target research area based on the spatial partitioning constraints, and generate several initial aggregation service range units with basic attributes. The indicator calculation module is used to map the features in the spatial feature dataset to the initial aggregation service range units and calculate the statistical indicators of each initial aggregation service range unit. The graph construction module is used to construct a graph structure with the initial aggregation service range units as nodes and the spatial adjacency relationship and semantic similarity between units as edges; The model building module is used to build a graph structure aggregation model based on the graph structure. The graph structure aggregation model includes an objective function aimed at minimizing the variance of statistical indicators among the initial aggregation service range units, and constraints based on maintaining the integrity of the spatial partitioning constraints. The optimization solution module is used to iteratively solve and partition the graph structure aggregation model using a heuristic search algorithm or a graph partitioning algorithm. Under the spatial partitioning constraints, it merges or reorganizes adjacent initial aggregation service range units to generate dynamic aggregation service range results that satisfy the statistical equilibrium constraints.

Images