An intelligent push processing method based on GIS city data mining

Abstract

The present application relates to the technical field of intelligent push processing based on GIS city data mining, and discloses an intelligent push processing method based on GIS city data mining. The complete process from data collection, structure construction to hot spot identification and push is covered. By setting the research area and grid division parameters, calculating the mean and standard deviation of the observation value, setting the quadtree decomposition threshold and maximum depth, constructing the root node and performing recursive decomposition, the leaf node set is finally formed. The hot spot candidate nodes meeting the level requirement are screened, the spatially adjacent hot spots are merged based on the adjacency matrix, and the push list is generated according to the priority. The method realizes spatial adaptive division through quadtree, improves the identification accuracy by combining statistical methods, enhances the hot spot aggregation ability by using connectivity analysis, and constructs a closed-loop system from raw data to intelligent push.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent push processing based on GIS city data mining, in particular to an intelligent push processing method based on GIS city data mining. BACKGROUND

[0002] With the accelerating process of urban informatization, city data mining and intelligent push technology based on geographic information system (GIS) have become an important part of smart city construction. By analyzing the spatio-temporal large-scale data of urban traffic flow and crowd activities, dynamic monitoring and precise service of hot areas can be realized, and the efficiency of urban management and public service can be improved.

[0003] In the prior art, intelligent push systems mostly rely on rule triggering or static threshold methods. The traditional method usually divides the city area into equal-area grids, aggregates the flow or event statistics of each grid globally, and then directly processes all grids according to fixed segmentation thresholds or preset empirical parameters. This approach ignores the differences in data distribution of different regions, making it difficult to achieve fine real-time detection in high fluctuation areas, often resulting in "over-pushing" or "missing detection" situations, affecting the accuracy and timeliness of hot area identification. In order to improve the spatial focusing ability, some researches try to apply spatial hierarchical indexing structures such as quadtree to city data analysis. Quadtree can achieve fine subdivision of high-density areas by continuously dividing the space twice, but existing solutions often use manually set segmentation thresholds and fixed maximum depth, lacking adaptive adjustment of segmentation criteria. In addition, the merging of leaf nodes also relies on global predefined parameters, which cannot be dynamically optimized according to the actual data correlation between nodes, leading to hot areas being easily divided into isolated small fragments or over-merged, reducing detection effectiveness. In terms of region merging and connectivity processing, the commonly used methods include clustering algorithms based on Euclidean distance or DBSCAN, which first group grids or small fragments according to density or distance thresholds, and then complete merging through connected component search in graph theory. This kind of method is highly sensitive to distance threshold and minimum neighbor number parameters, which need to be optimized in different scenarios, and the merging result is easily affected by noise data, making it difficult to ensure the authenticity and stability of spatial aggregation. In addition, existing technologies usually introduce empirical parameters in spatial division, threshold setting, decomposition depth and merging strategy, lacking a unified adaptive strategy, making the entire hot spot detection and intelligent push process too dependent on human experience, and unable to achieve end-to-end closed-loop automation in complex urban environments with multiple sources, multiple scales and multiple time periods. Especially in typical scenarios such as traffic peak, sudden activities or holidays, it is difficult to balance the real-time, accuracy and computational overhead.

[0004] To this end, the case aims to propose an intelligent push processing method based on GIS urban data mining, through setting reasonable spatial division and time window, using quadtree decomposition algorithm to recursively refine the spatial area, hierarchically expressing the dynamic data (such as taxi GPS point) in the city according to volatility, and then identifying the high-intensity hot spot area. Subsequently, the adjacent hot spots are merged by using the connected component search method, and the priority is sorted based on the statistical index, and finally the hot spot area list is output for intelligent push. SUMMARY

[0005] The application provides an intelligent push processing method based on GIS urban data mining, which promotes the solution to the problems mentioned in the background art.

[0006] The application provides the following technical scheme: an intelligent push processing method based on GIS urban data mining, comprising:

[0007] Setting the research area and the initial grid division parameter, constructing the grid set and determining the observation index;

[0008] Calculate the mean and standard deviation of the observation value of all grids, and set the quadtree decomposition threshold and the maximum decomposition depth;

[0009] Construct the root node of the quadtree, set the region boundary, the observation mean and the standard deviation;

[0010] Based on the set threshold and the observation data, the node is recursively decomposed, the child node is generated and the index and the statistical quantity are updated, until the termination condition is met;

[0011] Collect all nodes that no longer continue to decompose to form a leaf node set;

[0012] Screen the leaf nodes meeting the preset level condition as hot spot candidates;

[0013] Based on the adjacency matrix, the connected component search is carried out to merge the adjacent hot spot candidate nodes in space;

[0014] The priority of the merged hot spot area is sorted and the push list is generated.

[0015] Optionally, the setting of the research area and the initial grid division parameter, the construction of the grid set and the determination of the observation index specifically comprises:

[0016] The research area is a square area , and the side length is ;

[0017] The minimum leaf node side length is , and ;

[0018] The initial grid row and column number is , then the actual grid length is ;

[0019] Divide into equal-area grids to form a grid set :

[0020] , ; wherein, is the th grid; is the total number of grids;

[0021] Set the row number , column number for each , and the area is: ; wherein, is the two-dimensional coordinate range of the grid ; is the modulo operation;

[0022] Set the time window to the past 1 hour;

[0023] In the time window, get the total number of times the taxi GPS device reports positioning points in the grid area, denoted as the observation value .

[0024] Optionally, the mean and standard deviation of the observation values of all grids are calculated, and a quadtree decomposition threshold and a maximum decomposition depth are set, specifically including:

[0025] Calculate the arithmetic mean of all grid observation values , ;

[0026] Calculate the global standard deviation , ;

[0027] Set the quadtree decomposition threshold to ;

[0028] Set the maximum quadtree decomposition depth to .

[0029] Optionally, the root node of the quadtree is constructed, and the area boundary, observation mean and standard deviation are set, specifically including:

[0030] Set the quadtree root node to , ; wherein, is the depth of the node in the tree, and the root node ;

[0031] set the root node region boundary as , ; wherein, is the left boundary coordinate of the node ; is the right boundary coordinate of the node ; is the upper boundary coordinate of the node ; is the root node;

[0032] set the root node grid index set as , ; wherein, is the number of grids of the node ; is the set containing all grid indexes within the region ;

[0033] set the observation mean and observation standard deviation within the root node region as , ; wherein, is the observation mean within the region of the node ; is the observation standard deviation within the region of the node.

[0034] Optionally, the recursive decomposition of the node based on the set threshold and observation data, the generation of the child node and the update of the index and statistical quantity are performed until a termination condition is met, and specifically include:

[0035] for any node , initially set , and the following steps are performed:

[0036] S101, determine whether to decompose:

[0037] if , decompose; otherwise, mark as a leaf node and return;

[0038] S102, calculate the node midpoint coordinate:

[0039] , ; wherein, is the center point coordinate of the region of the current node ;

[0040] S103, generate four child nodes:

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] wherein, is the i-th child node of node ;

[0046] and let ;

[0047] S104, child node grid index:

[0048] recalculate , , ; wherein, is the number of elements in the combination;

[0049] if , discard the child node; otherwise, continue;

[0050] S105, child node statistics:

[0051] , ;

[0052] for each repeat steps S101 to S105 for the child nodes of .

[0053] Optionally, the collection of all nodes that no longer continue to decompose forms a set of leaf nodes, specifically including:

[0054] collecting all nodes that no longer continue to decompose as leaf nodes:

[0055] ; wherein, is the set of all nodes that no longer decompose.

[0056] Optionally, the leaf nodes that meet the preset level condition are selected as hotspot candidates, specifically including:

[0057] selecting nodes that reach the maximum level from the leaf nodes as hotspot candidates:

[0058] ; wherein, is the set of hotspot candidate nodes.

[0059] Optionally, the connected component search is performed based on the adjacency matrix, and spatially adjacent hotspot candidate nodes are merged, specifically including:

[0060] setting a matrix , wherein ​For any candidate node :

[0061] ;

[0062] wherein, is a matrix element;

[0063] Connected component search:

[0064] S210, initialize an unvisited set , a connected component list ;

[0065] S220, when :

[0066] S221, take a node from the unvisited set , set the current component set , the queue , and remove from the unvisited set ;

[0067] S222, when , pop the node ; for each if , add to the current component set and , and remove from the queue ; wherein, is the current node and adjacent nodes in the traversal process;

[0068] S223, add the completed component to the connected component list ;

[0069] S230, end ;

[0070] For each component , respectively calculate the range of the first merged hotspot area and the set of all grid indexes contained in the area :

[0071] , .

[0072] Optionally, the priority sorting and generating the push list for the merged hotspot area, specifically includes:

[0073] calculate the first Mean of hotspot area Standard deviation :

[0074] , ; wherein, is the cardinality of the set ;

[0075] Arrange all the merged regions from large to small; Select the first

[0076] region , output the push tuple:

[0077] , .

[0078] The present application has the following beneficial effects:

[0079] 1. The dynamic setting method of the minimum leaf node side length and the number of grid rows and columns is introduced, so that the space division has good scalability and adaptability; the time window setting is combined with the space division, so that the system can realize the space-time synchronous analysis; the clear two-dimensional coordinate conversion and index mechanism provides a unified space reference for subsequent algorithm processing. The traditional heat map method mainly uses static grid division, which cannot automatically adjust the region granularity according to the spatial density of data. The present scheme improves the spatial expression ability through parameter setting and automatic grid numbering, and provides a solid foundation for subsequent data aggregation and decomposition.

[0080] 2. The global mean and standard deviation are used to dynamically calculate the decomposition threshold, so that the decomposition criterion is driven by data; the maximum decomposition depth control mechanism is proposed, which effectively avoids the waste of computing resources caused by infinite refinement of the quadtree; it has strong generalization ability and is suitable for different regional and different city scale data analysis scenes. Compared with the static threshold mechanism in the prior art which needs to be set by artificial experience, the present scheme automatically analyzes the statistical quantity to adaptively judge the necessity of region decomposition, which significantly improves the accuracy and efficiency of hot area identification.

[0081] 3. A unified space management structure starting from the whole region is established, so that the region division and node management have recursive traceability; all grids and root node index mapping are clear, which is convenient for structure operations such as region merging and region searching; the initial statistical value is driven by actual data, which lays a reasonable decomposition initial condition. The traditional method does not have a special structure to unify the scheduling of space nodes, resulting in complicated subsequent operations. The present method encapsulates the initial state through the root node, which improves the organization and controllability of the space tree structure.

[0082] ​4. Adopt hierarchical recursive structure, refine region level by level, realize hierarchical detection of spatial hotspots from coarse to fine; each level of child node recalculates local statistical value, adaptively judges whether to continue decomposition; discard invalid nodes without grid, optimize data structure for simplicity and memory occupation. Compared with traditional static grid analysis, this method controls through "meaningful decomposition + termination mechanism", not only avoids redundant region processing, but also improves spatial resolution, realizes accurate positioning of hotspots.

[0083] 5. Extract all nodes that are no longer refined as "leaf nodes", constitute the basic set for subsequent analysis; filter according to hierarchical conditions, only keep the most fine-grained and most significant fluctuation region as the hotspot candidate. The prior art generally uses fixed threshold or manual determination of hotspot position, lacking spatial hierarchical sensitivity. This scheme is based on the screening mechanism combining decomposition level and data fluctuation, which is more scientific and reasonable.

[0084] 6. Construct an adjacency matrix, use connected component search to merge adjacent hotspot nodes, fully retain the continuity of hotspots; use breadth-first search algorithm of graph structure to efficiently cluster, ensure the coherence and integrity of region merging; output the merged spatial range and grid index set, facilitate subsequent comprehensive evaluation. Unlike traditional methods based only on numerical merging, this scheme emphasizes the dual merging standard of "spatial proximity + numerical significance", improving the integrity of hotspot region representation and the fidelity of urban spatial features.

[0085] 7. For each hotspot region, calculate its statistical indicators (mean, standard deviation), establish a unified sorting rule; output the hotspot region with the highest priority to push the tuple, realize the closed loop from data perception to action instruction; support real-time push service deployment, such as police patrol, emergency response, etc. BRIEF DESCRIPTION OF DRAWINGS

[0086] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0087] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0088] Embodiment, refer to Figure 1 A smart push processing method based on GIS urban data mining, comprising:

[0089] Set the research area and initial grid division parameters, construct the grid set and determine the observation index;

[0090] Compute the mean and standard deviation of the observation values of all grids, and set the threshold and maximum decomposition depth of the quadtree decomposition;

[0091] Construct the root node of the quadtree, set the region boundary, observation mean and standard deviation;

[0092] Based on the set threshold and observation data, recursively decompose the node to generate child nodes and update the index and statistics until the termination condition is met;

[0093] Collect all nodes that no longer continue to decompose to form a leaf node set;

[0094] Screen the leaf nodes that meet the preset level conditions as hotspot candidates;

[0095] Based on the adjacency matrix, search for connected components and merge spatially adjacent hotspot candidate nodes;

[0096] Prioritize the merged hotspot regions and generate a push list.

[0097] A complete and intelligent push processing method for GIS city data is proposed, covering the whole process from data collection, structure organization to hotspot identification, priority sorting and push. By introducing recursive quadtree structure, the city spatial data is hierarchically organized and decomposed, effectively dealing with the problem of different regional hotspot density differences; then by setting the decomposition threshold and priority evaluation index through statistical method, the accuracy of hotspot identification is improved. Finally, through the adjacency matrix, the spatially adjacent hotspot regions are merged, realizing the closed-loop process from "raw data" to "high-quality hotspot push". By setting the spatial division parameters and combining the spatio-temporal observation data, the problem of not timely and accurate capturing of city hotspot changes is solved; through recursive decomposition and connected clustering, multi-scale hotspot regions are automatically discovered, avoiding the subjectivity caused by manual zoning; the push mechanism improves the response speed and accuracy, and is widely applicable to city safety warning, traffic scheduling, emergency management and other scenarios.

[0098] The set research area and initial grid division parameters are constructed, and the grid set is determined and the observation index is determined, specifically including:

[0099] The research area is a square area with a side length of ; the overall spatial range is clear, providing a geometric reference for grid division;

[0100] The minimum leaf node side length is set to , and ; the minimum division granularity is controlled to balance the resolution and computational complexity;

[0101] The initial grid row and column number is set to , then the actual grid side length is ; determine the number of rows, columns and actual length of grid division, ensure the complete coverage of the area;

[0102] Divide into equal-area grids to form a grid set :

[0103] , ; wherein, is the grid; is the total number of grids; an index set of all grid units is established to provide a unified identification for subsequent reference;

[0104] Set the row number , column number for each , and its area is: ; wherein, is the two-dimensional coordinate range of the grid ; is the modulo operation; map each grid to the actual plane coordinate for subsequent point-area attribution judgment;

[0105] Set the time window to the past 1 hour;

[0106] In the time window, get the total number of taxi GPS device positioning points reported in the grid area, denoted as the observed value .

[0107] Around "setting the research area and grid division and observation index acquisition", it is the basic step of the whole data mining process. By defining the city area as a standard square and setting the minimum leaf node length and the initial grid row and column number according to the actual scene, a unified grid system is established, and the taxi GPS data is taken as the observation object to realize the real-time quantification of the city operation status. Through this step, the unified digital modeling of the city space is realized, so that the subsequent analysis has a spatial reference standard; the grid processing makes the original unstructured point data become operable two-dimensional matrix data, enhancing the scalability and processing efficiency of data analysis; after setting the time window, the data analysis has timeliness, improving the adaptability of the system in dynamic scenarios. These settings lay a solid spatio-temporal foundation for subsequent quadtree decomposition and hotspot identification.

[0108] The average value and standard deviation of the observation value of all grids are calculated, and the quadtree decomposition threshold and maximum decomposition depth are set, which specifically includes:

[0109] Calculate the arithmetic mean of all grid observation values , ; provide the centralized trend benchmark of the whole area data;

[0110] Calculate the global standard deviation , ; quantify the dispersion degree of the whole area data, and provide a quantitative basis for the decomposition threshold;

[0111] Set the quadtree decomposition threshold to ; adaptively use the global fluctuation as the criterion for further decomposition, without external parameter adjustment;

[0112] Set the maximum quadtree decomposition depth to ; limit the decomposition depth to prevent excessive refinement and cause computational explosion.

[0113] Focusing on "global statistical quantity calculation and decomposition parameter setting" is the key link for the scientific development of quadtree decomposition. It first calculates the mean and standard deviation of all grid observations, providing a benchmark for the overall data distribution of the region. Then, it sets the quadtree decomposition threshold and maximum decomposition level based on the standard deviation, determining when to decompose and when to terminate. By data-driven setting of the decomposition threshold, the subjectivity and inadaptability of manual setting in traditional methods are overcome, realizing the adaptive ability of the model. By setting the maximum decomposition depth, the computational cost is effectively controlled, ensuring the stable operation of the system. This step solves the problems of strong randomness of decomposition strategy, difficulty in balancing precision and resource consumption, and is the core prerequisite for fine-grained hotspot identification.

[0114] The root node of the quadtree is constructed, and the region boundary, observation mean and standard deviation are set, which specifically includes:

[0115] Set the root node of the quadtree as , ; where is the depth of node in the tree, and the root node ; the starting point of constructing the quadtree, and the initial level is marked as 0;

[0116] Set the root node region boundary as , ; where is the coordinate of the left boundary of node ; is the coordinate of the upper boundary of node , here is the root node; clearly define the complete geometric area covered by the root node;

[0117] Set the root node grid index set as , ; where is the number of grids; a set of grid indices contained in the region ;

[0118] set the observed mean and observed standard deviation in the root node region as , ; wherein, is the observed mean in the node region; is the observed standard deviation in the node region;

[0119] establish a mapping of the root node and all grids, and calculate initial statistics for decomposition determination.

[0120] The process of "building a quadtree root node" is described. By setting the initial depth of the node, the region boundary and the set of contained grid indices, and calculating the observed mean and standard deviation, a starting point is provided for the establishment of the entire quadtree recursive structure. The root node uniformly carries the spatial range, data index and initial statistical indicators, making the subsequent node decomposition directional and targeted; the unified index system and boundary definition lay the foundation for recursive operation and hotspot area positioning; in addition, the initial statistical information of the root node is also used to determine whether the first round needs to be decomposed, thereby improving processing efficiency and avoiding the redundant processing of the entire region as a whole.

[0121] The node is recursively decomposed based on the set threshold and observation data, sub-nodes are generated and the index and statistics are updated until the termination condition is met, specifically including:

[0122] for any node , initially , the following steps are performed:

[0123] S101, determine whether to decompose:

[0124] if , decompose; otherwise, mark as a leaf node and return; only nodes with large fluctuations and not reaching the maximum level are decomposed to ensure meaningful refinement;

[0125] S102, calculate the node midpoint coordinates:

[0126] , ; wherein, is the center point coordinate of the current node region; determine the center of the current node region for equal division of sub-regions;

[0127] S103, generate four sub-nodes:

[0128] ;

[0129] ;

[0130] ;

[0131] ;

[0132] wherein, is the i-th child node of node ;

[0133] and let ;

[0134] Split the region by the midpoint to achieve the quadtree decomposition;

[0135] S104, child node grid index:

[0136] Recalculate , , ; wherein, is the number of elements combined in the middle; determine the grid set responsible for each child node, prepare for local statistics;

[0137] If , discard the child node; otherwise, continue;

[0138] S105, child node statistics:

[0139] , ; obtain the mean and volatility of the sub-region, used for the next round of decomposition decision or leaf node confirmation;

[0140] For each , repeat steps S101 to S105; continue iteration until all high volatility regions are refined to the minimum level or meet the uniformity.

[0141] It judges whether each node needs to be decomposed layer by layer according to the set decomposition threshold and maximum level. If the condition is met, the region is divided into four child nodes, and the index and statistics of each child node are updated until all nodes meet the "non-decomposition" condition. By introducing a dynamic judgment mechanism, only regions with spatial data fluctuations are decomposed, reducing unnecessary calculations; continuously iterating and optimizing the region granularity, so that the final hotspot recognition can achieve the most reasonable spatial resolution; through the statistics value driven decomposition logic, the accuracy and intelligent level of region processing are greatly enhanced, adapting to the needs of large-scale data processing.

[0142] The collection of all nodes that no longer continue to decompose forms a leaf node set, specifically including: ​

[0143] Collect all nodes that are not further decomposed as leaf nodes:

[0144] wherein, is the collection of all leaf nodes that are not further decomposed;

[0145] Collect all nodes that terminate decomposition to form a base collection for subsequent hotspot screening.

[0146] The collection of leaf nodes formed by collecting all "terminate decomposition" nodes is a prerequisite for hotspot preliminary screening. Leaf nodes are the terminal nodes of quadtree decomposition, representing the most granular regions of information. This step unifies the terminal regions in the early decomposition process to form a candidate collection for the next step of hotspot identification. Through focused analysis of these regions, the accuracy and efficiency of hotspot detection can be greatly improved. At the same time, the system data flow is clear, and it has a good structured foundation, which is convenient for subsequent screening and sorting operations.

[0147] The leaf nodes that meet the preset level condition are selected as hotspot candidates, specifically including:

[0148] Select nodes that reach the maximum level from the leaf nodes as hotspot candidates:

[0149] wherein, is the collection of hotspot candidate nodes;

[0150] Select regions with the finest granularity and still high fluctuations as hotspot preliminary candidates.

[0151] On the basis of leaf nodes, further filter nodes that meet the maximum level as hotspot candidates, excluding nodes with low levels or small data fluctuations. Through the level filtering mechanism, focus on the data units that are decomposed to the finest region and have the most significant fluctuations, achieving precise positioning of abnormal regions or sudden events; avoid misjudging slight data fluctuations in larger regions as hotspots, improving judgment quality; further compressing the scale of processing data, improving the efficiency of subsequent clustering and sorting.

[0152] The connected component search based on the adjacency matrix merges spatially adjacent hotspot candidate nodes, specifically including:

[0153] Set the matrix wherein for any candidate node :

[0154] ;

[0155] wherein, is the element of the matrix ;

[0156] Connected component search:

[0157] S210, initialize the unvisited set , connected component list ;

[0158] S220, when :

[0159] S221, take a node from the unvisited set , set the current component set , the queue , and remove from the unvisited set ;

[0160] S222, when , pop the node ; for each if , add to the current component and , and remove from the queue ; wherein is the current node and adjacent nodes in the traversal process

[0161] S223, add the completed component to the connected component list ;

[0162] S230, end when ;

[0163] By the queue traversal algorithm, aggregate all adjacent hot spot nodes

[0164] For each component , respectively calculate the first merged hot spot area set range and all grid index sets contained in the area :

[0165] , ;

[0166] Merge all child node spaces within the component, and summarize the contained grids, to prepare for overall hot spot statistics.

[0167] Through the adjacency matrix and connected component search algorithm, the spatially continuous hot spot candidate nodes are merged to generate a hot spot area set with stronger structure and higher expression ability. Through the graph structure traversal method, the spatial connectivity between hot spots is fully considered to avoid fragmentation judgment. At the same time, multiple small hot spots with dense distribution are merged into a unified area to improve the strategy and concentration of information pushing. Compared with the traditional threshold-based simple merging method, it has more spatial intelligence and algorithm robustness, and is especially suitable for complex spatial patterns in cities.

[0168] The merged hot spot area is prioritized and a push list is generated, specifically including:

[0169] The mean and standard deviation of the first hot spot area are calculated:

[0170] , ; wherein, is the cardinality of the set ; the overall volatility of each hot spot area is evaluated as a subsequent priority basis;

[0171] All merged areas are arranged in descending order according to ; the area with the largest volatility is placed in the highest priority;

[0172] The first areas are selected, and the push tuple is output:

[0173] , ;

[0174] The final intelligent push instruction is formed, and the hot spot detection-pushing process is closed loop.

[0175] The mean and standard deviation of each hot spot area are calculated, the hot spot areas are sorted, and a number of high-priority areas are selected to form the final push list. The sorting mechanism introduces a quantitative evaluation system, so that the hot spot pushing is not only based on location, but also combines the comparability of data strength; the push list structure is clear, which is convenient for direct application by various management systems; a closed-loop management mode from data mining to task allocation is realized, which greatly improves the intelligent level and practicality of the system, and has wide application prospects in city management, traffic emergency, environmental warning and other fields.

[0176] It is to be noted that, in the present text, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0177] The above description is merely preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, some improvements and refinements can be made without departing from the technical principles of the present application, and these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. A smart push processing method based on GIS urban data mining, characterized in that, include: Define the study area and initial grid division parameters, construct the grid set, and determine the observation indicators; The process of setting the study area and initial grid division parameters, constructing the grid set, and determining the observation indicators specifically includes: Assume the study area is a square region. The side length is ; Let the side length of the smallest leaf node be... ,and ; Let the initial number of raster rows and columns be... The actual grid side length is ; Will Average score A grid of equal area forms a grid set. : , ;in, For the first One grid; This represents the total number of grid cells. For each Set line numbers , column number Its area is: ;in, For grid The range of two-dimensional coordinates; This is the modulo operation; Set the time window to the past hour; Within the time window, acquire the taxi GPS device's position on the grid. The total number of times the location points were reported within the area is recorded as the observation value. ; Calculate the mean and standard deviation of observations for all rasters, and set the quadtree decomposition threshold and maximum decomposition depth; The calculation of the mean and standard deviation of all raster observations, and the setting of the quadtree decomposition threshold and maximum decomposition depth, specifically includes: Calculate the arithmetic mean of all raster observations. , ; Calculate the global standard deviation , ; Set the quadtree decomposition threshold to 0. ; Set the maximum decomposition depth of the quadtree to ; Construct the root node of the quadtree, and set the region boundaries, observation mean, and standard deviation; The construction of the root node of the quadtree, setting the region boundary, observation mean, and standard deviation, specifically includes: Let the root node of the quadtree be , ;in, For nodes Depth in the tree, root node ; Let the boundary of the root node region be , ;in, For nodes left and right boundaries coordinate; For nodes upper and lower boundaries Coordinates, here It is the root node; Set the root node raster index set to , ;in, For nodes Grid number; For included region The set of all raster indexes within; Set the mean and standard deviation of the observations within the root node region as follows: , ;in, For nodes The average value observed within the region; The standard deviation of observations within the node region; Based on a set threshold and observation data, nodes are recursively decomposed to generate child nodes and update indexes and statistics until the termination condition is met. The recursive decomposition of nodes based on a set threshold and observation data, generating child nodes and updating indexes and statistics, continues until a termination condition is met. Specifically, this includes: For any node Initially Perform the following steps: S101. Determine whether to decompose: like If it is true, then decompose it; otherwise, mark it as a leaf node and return. S102. Calculate the coordinates of the midpoint of the node: , ;in, For the current node The coordinates of the center point of the region; S103. Generate four child nodes: ； ； ； ； in, For nodes The Number of child nodes; And order ; S104, Child Node Raster Index: For each Recalculate , ;in, The number of elements in the combination; like If the condition is met, discard the child node; otherwise, continue. S105, Child Node Statistics: ， ； For each make Repeat steps S101 to S105 for the child nodes; Collect all nodes that are no longer decomposed, forming a set of leaf nodes; The process of collecting all nodes that are no longer decomposed, forming a leaf node set, specifically includes: Collect all nodes that have not been further decomposed into leaf nodes: ;in, This is the set of all leaf nodes that are no longer decomposed. Leaf nodes that meet the preset hierarchical conditions are selected as hotspot candidates; The process of selecting leaf nodes that meet preset hierarchical conditions as hotspot candidates specifically includes: Select nodes that reach the maximum level from the leaf nodes as hotspot candidates: ;in, A set of hotspot candidate nodes; Connectivity component search is performed based on the adjacency matrix, and spatially adjacent hotspot candidate nodes are merged. The connected component search based on the adjacency matrix, merging spatially adjacent hotspot candidate nodes, specifically includes: Set matrix ,in For any candidate node : ； in, For matrix element; Connected component search: S210, Initialize the unaccessed set Connected component list ; S220, when : S221, from Take a node Let the current component set be... ,queue and will from Remove from; S222, when pop up nodes For each like Then join in and and from Remove from ;in, This refers to the current node and its adjacent nodes during the traversal process; S223, The completed component join in ; S230, End ; For each component Calculate the first one respectively The range of the merged hotspot areas With the set of all raster indexes contained in this region : ， ； Prioritize the merged hotspot areas and generate a push notification list.

2. The intelligent push processing method based on GIS urban data mining according to claim 1, characterized in that, The process of prioritizing the merged hotspot areas and generating a push list specifically includes: Calculate the first The average of hotspot areas with standard deviation : , ;in, For set The cardinality; according to Arrange all merged regions from largest to smallest; Before selection each region Output push tuple: ， 。

Images