Geographic information retrieval method for grid management of personal loan non-performing assets based on spatial index

Abstract

The invention relates to the technical field of data processing, and discloses a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial index. The method comprises the following steps: performing initial grid division on a target area to form a first grid set, calculating the density of non-performing assets of each grid to obtain a first density distribution; judging whether a first condition for dynamic grid adjustment is triggered according to the first density distribution, and if triggered, performing adaptive re-division to generate a second grid set; constructing a second spatial index tree, receiving a retrieval request, and performing retrieval according to the second spatial index tree to generate a retrieval result; judging whether a second condition for dynamic grid adjustment is triggered according to the retrieval result, and if triggered, dynamically adjusting grid parameters and re-dividing, and if not triggered, outputting the retrieval result; the invention realizes efficient and accurate retrieval of individual loan non-performing assets by dynamically adjusting grid division and spatial index tree, and greatly improves the flexibility and retrieval efficiency of geographic information management.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and more specifically, to a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing. Background Art

[0002] At present, there are some related technologies for spatial data management and query. For example, the Chinese patent application with publication number CN116932610A discloses an efficient outlier query processing method based on grid-tree partitioning. The method obtains an ordered grid set by establishing a hierarchical grid of the data set, and performs VG tree partitioning according to the ordered grid set. Then, based on the formed 1-K distance table as an index, the real outliers of the data set are obtained given the number of query neighbors and the number of outliers. The method can query outliers quickly and accurately. However, the method is mainly aimed at static data sets, lacks support for dynamically changing data, and cannot adapt to the dynamic characteristics of non-performing personal loan assets. In addition, the Chinese patent application with publication number CN116934968A discloses a three-dimensional visualization method for network security assets. The method generates a three-dimensional model of the asset in a spatial coordinate system based on the vector data and geographic location data of the asset, and uses a quadtree spatial index to divide the three-dimensional model into nodes. By binding the asset symbol with the three-dimensional model, the three-dimensional visualization expression of security assets in the digital twin environment is realized. However, this method mainly focuses on the three-dimensional visualization of assets, and its ability to retrieve geographic information of assets is relatively weak.

[0003] In summary, most existing technologies perform spatial indexing and query processing on static data sets and lack effective support for dynamically changing data; at the same time, the retrieval flexibility and efficiency of existing methods need to be further improved. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a geographic information retrieval method for grid management of personal loan non-performing assets based on spatial index. In view of the characteristics of personal loan non-performing assets such as wide geographical distribution, large total amount, small single amount and dynamic changes, the present invention can dynamically adjust the grid division and spatial index, and provide a geographic information management method for efficient and accurate retrieval. By dynamically adjusting the grid division and spatial index tree, efficient and accurate retrieval of personal loan non-performing assets can be achieved, greatly improving the flexibility of geographic information management and retrieval efficiency.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] Performing initial grid division on the target area to form a first grid set, calculating the non-performing asset density of each grid in the first grid set to obtain a first density distribution; judging whether a first condition for dynamic grid adjustment is triggered according to the first density distribution; if triggered, performing adaptive re-division on the first grid set to generate a second grid set; if not triggered, maintaining the first grid set;

[0007] For the first grid set, construct a first spatial index tree; according to the first grid set and the second grid set, obtain an incremental grid set; according to the incremental grid set, construct an incremental spatial index tree; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree;

[0008] Receive a geographic information retrieval request, perform retrieval according to the second spatial index tree, and generate retrieval results; determine whether the second condition of dynamic grid adjustment is triggered according to the retrieval results; if triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-divide the target area into initial grids; if not triggered, complete this retrieval and output the retrieval results.

[0009] Furthermore, the initial grid division of the target area to form a first grid set includes:

[0010] Obtain the geographic location information of individual loan non-performing assets and unify the geographic location information into longitude and latitude coordinates in the WGS84 coordinate system;

[0011] According to the preset first grid size parameter and the first grid shape parameter, the target area is initially grid-divided to form n1 grids; n1 is a positive integer;

[0012] Extracting core attribute information of each grid; the core attribute information includes grid area, grid quantity, grid number, center coordinates and boundary range;

[0013] Count the number of individual loan non-performing assets that fall into each grid, obtain the ID list of these individual loan non-performing assets, and establish a grid-asset mapping index;

[0014] A first grid set is formed by core attribute information and grid-asset mapping index.

[0015] Further, the calculation of the non-performing asset density of each grid in the first grid set to obtain the first density distribution includes: traversing each grid in the first grid set, calculating the non-performing asset density of each grid by the ratio of the number of non-performing assets in the grid to the grid area, generating the first density distribution, and obtaining the distribution vector of the first density distribution. ; is the non-performing asset density of the ith grid, is the grid number in the first grid set, ;

[0016] The first condition for determining whether to trigger dynamic grid adjustment according to the first density distribution includes: determining whether to trigger dynamic grid adjustment according to the first density distribution ... , calculating the discrete degree of the first density distribution; comparing the discrete degree of the first density distribution with a preset discrete threshold, and triggering the first condition for dynamic grid adjustment if the discrete degree of the first density distribution exceeds the discrete threshold.

[0017] Further, the performing adaptive re-division on the first grid set to generate the second grid set includes:

[0018] Calculate the upper quartile and the lower quartile of the first density distribution, use the upper quartile as the first density threshold of the first density distribution, and use the lower quartile as the second density threshold of the first density distribution;

[0019] Traversing the first grid set, dividing the grids whose non-performing asset density is higher than the first density threshold into four equal parts, and dividing them into four sub-grids to form a first sub-grid set;

[0020] Traversing the first grid set, merging the grids whose non-performing asset density is lower than the second density threshold, and merging two grids into one to form a mother grid, thereby obtaining a first mother grid set;

[0021] The first sub-grid set, the first mother grid set and the grids in the first grid set that have not been adaptively re-divided are integrated to obtain an adjusted grid set, and all grids in the adjusted grid set are numbered to form a second grid set.

[0022] Furthermore, the method for numbering all grids in the adjusted grid set includes:

[0023] Mark the grids in the first grid set that performs the quartering as the quartered original grids, hide the quartered original grids, and add two subscripts to the numbers of the four newly added subgrids on the basis of the numbers of the quartered original grids to obtain the grid numbers of all the subgrids in the first subgrid set;

[0024] For each mother grid formed by combining two grids into one, the grid number of all mother grids in the first mother grid set is obtained by taking the grid number with the smaller lexicographic order of the original numbers of the two merged grids;

[0025] For the grids in the first grid set that do not perform adaptive re-division, their numbers remain unchanged;

[0026] Summarize the grid numbers of all subgrids in the first subgrid set, the grid numbers of all mother grids in the first mother grid set, and the grid numbers of grids that have not been adaptively repartitioned in the first grid set to obtain the grid numbers of all grids in the adjusted grid set;

[0027] All grids in the adjusted grid set are renumbered according to the lexicographic order of the grid numbers; the lexicographic order of the grid numbers includes an increasing lexicographic order and a decreasing lexicographic order.

[0028] Further, obtaining the incremental grid set according to the first grid set and the second grid set includes:

[0029] Traversing the second grid set, extracting grids whose numbers do not exist in the first grid set, to form a newly added grid set;

[0030] Traversing the first grid set, extracting grids whose numbers do not exist in the second grid set, to form a deleted grid set;

[0031] Traversing the second grid set, extracting the grids whose numbers are the same as those of a grid in the first grid set but whose core attribute information has changed, to form an updated grid set;

[0032] The newly added grid set, deleted grid set, and updated grid set are unioned to obtain the incremental grid set.

[0033] Furthermore, the geographic information search request includes the geographic location of the target individual loan non-performing assets, the target search range parameter and the target grid granularity parameter;

[0034] The performing of the search and generating the search results comprises:

[0035] Extracting a target grid granularity parameter as a second grid granularity parameter, comparing the second grid granularity parameter with the first grid granularity parameter, and determining whether the second grid granularity parameter is consistent with the first grid granularity parameter;

[0036] If the second grid granularity parameter is consistent with the first grid granularity parameter, the search is directly performed using the second spatial index;

[0037] If the second grid granularity parameter is inconsistent with the first grid granularity parameter, the second grid set is re-divided according to the second grid granularity parameter to form a third grid set, and a third spatial index tree corresponding to the third grid set is established, and the third spatial index tree is used to perform retrieval.

[0038] Furthermore, the performing of the search and generating the search results further includes:

[0039] According to the target search range parameters, determine the spatial boundary of the rectangular search area centered on the geographical location of the target non-performing assets;

[0040] Calculate the grid numbers of the starting grid and the ending grid covered by the rectangular search area, and use them as the search starting grid number and the search ending grid number respectively;

[0041] Using the second spatial index tree or the third spatial index tree, traverse the grid set covered by the search start grid number to the search end grid number to form a candidate grid set;

[0042] The geographical information of non-performing assets contained in the candidate grid set is extracted to generate a search result.

[0043] Furthermore, the second condition for determining whether to trigger dynamic grid adjustment includes:

[0044] Perform cluster analysis on the search results to generate search quality evaluation parameters that reflect the rationality of grid division within the time and space range; the search quality evaluation parameters include cluster grid density dispersion coefficient GDDC, cluster asset density dispersion coefficient ADDC, intra-cluster grid granularity variation coefficient GGVC, and inter-cluster grid granularity variation coefficient GGVB;

[0045] Set the grid density discreteness threshold for the clustering grid density discreteness coefficient GDDC ; Set the asset density discrete threshold for the cluster asset density discrete coefficient ADDC ; Set the intra-cluster grid size variation threshold for the intra-cluster grid size variation coefficient GGVC ; Set the inter-cluster grid size variation coefficient GGVB to set the inter-cluster grid size variation ;

[0046] When two or more retrieval quality evaluation parameters exceed their respective thresholds, a second condition for triggering dynamic grid adjustment is determined.

[0047] The spatial index-based individual loan non-performing assets grid management geographic information retrieval system is used to implement the spatial index-based individual loan non-performing assets grid management geographic information retrieval method, the system comprising:

[0048] Grid division module: used to perform initial grid division on the target area to form a first grid set, calculate the non-performing asset density of each grid in the first grid set, and obtain a first density distribution; according to the first density distribution, determine whether the first condition for dynamic grid adjustment is triggered; if triggered, perform adaptive re-division on the first grid set to generate a second grid set; if not triggered, keep the first grid set;

[0049] Spatial index module: used to construct a first spatial index tree for a first grid set; obtain an incremental grid set according to the first grid set and the second grid set; construct an incremental spatial index tree according to the incremental grid set; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree;

[0050] Retrieval module: used to receive geographic information retrieval requests, perform retrieval according to the second spatial index tree, and generate retrieval results; based on the retrieval results, determine whether the second condition for dynamic grid adjustment is triggered; if triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-initialize the grid division of the target area; if not triggered, complete this retrieval and output the retrieval results.

[0051] An electronic device includes a memory, a central processing unit, and a computer program stored in the memory and executable on the central processing unit. When the central processing unit executes the computer program, the above-mentioned spatial index-based individual loan non-performing asset grid management geographic information retrieval method is implemented.

[0052] A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the above-mentioned spatial index-based individual loan non-performing asset grid management geographic information retrieval method is implemented.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The present invention forms a multi-resolution, multi-granularity grid system by dynamically gridding the target area, and constructs an efficient spatial index tree based on the grid, so that the grid granularity and retrieval strategy can be flexibly adjusted according to different retrieval requirements, the target data can be quickly locked and screened, and the retrieval efficiency and management flexibility can be greatly improved.

[0055] By calculating the density of non-performing assets in each grid, the present invention can intuitively reflect the aggregation and distribution patterns of non-performing assets in geographical space; by analyzing the discrete degree of the non-performing asset density distribution, abnormal areas of spatial distribution can be quickly identified; by dynamically adjusting the grid granularity based on the density distribution, a finer-grained characterization can be achieved in dense areas, and a generalized expression can be achieved in sparse areas, thereby strengthening the management and control of high-risk areas in a targeted manner.

[0056] The present invention closely associates the geographic information and business attributes of individual loan non-performing assets with the grid management unit. By timely updating the index tree for newly added, deleted, and updated grids, the dynamic changes in the distribution of non-performing assets can be captured in real time. By analyzing the spatiotemporal evolution of density distribution, the diffusion trend of non-performing assets can be predicted. By evaluating retrieval quality feedback and optimizing grid parameters, the accuracy of identifying and reaching non-performing assets can be continuously improved, providing strong support for risk prevention, asset disposal, etc.

[0057] The present invention automatically determines whether to trigger dynamic grid adjustment according to the discrete degree of non-performing asset density distribution, introduces an adaptive re-division mechanism, and optimizes the grid structure by quartering, merging, etc. At the same time, based on the retrieval quality evaluation index, it identifies the rationality of grid division, further guides the dynamic optimization of grid granularity and shape parameters, realizes the coordinated evolution of the grid system and the distribution of non-performing assets, and continuously enhances the adaptability to the complex and changeable geographical distribution of non-performing assets.

[0058] When the distribution of non-performing assets changes, the present invention does not simply rebuild the entire grid set and index tree, but uses the old and new grid sets to identify the changed parts, and only constructs an incremental grid set for the newly added, deleted, and updated grids, and then generates an incremental index tree. Finally, the incremental index tree is merged with the original index tree, avoiding a large number of unnecessary repeated calculations, greatly reducing the time and space overhead of dynamic adjustment, and improving the real-time response speed of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1 It is a principle flow chart of the geographic information retrieval method for grid management of individual loan non-performing assets based on spatial index in the present invention;

[0061] Figure 2 It is a flow chart of a method for performing initial grid division on a target area in a geographic information retrieval method for grid management of non-performing assets of individual loans based on spatial indexes of the present invention;

[0062] Figure 3 A flowchart of a method for determining whether to trigger a first condition of dynamic grid adjustment in a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing of the present invention;

[0063] Figure 4A flow chart of a method for generating a second grid set in a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing according to the present invention;

[0064] Figure 5 A flowchart of a method for obtaining an incremental grid set in a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing of the present invention;

[0065] Figure 6 A flowchart of a method for constructing an incremental spatial index tree in a spatial index-based individual loan non-performing asset grid management geographic information retrieval method of the present invention;

[0066] Figure 7 A flowchart of a method for performing retrieval and generating retrieval results in a geographic information retrieval method for grid management of non-performing assets of individual loans based on spatial indexes according to the present invention;

[0067] Figure 8 A flowchart of a method for determining whether to trigger a second condition of dynamic grid adjustment in a geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing of the present invention;

[0068] Fig. 9 It is a functional module diagram of the geographic information retrieval system for grid management of individual loan non-performing assets based on spatial index in the present invention;

[0069] Fig.10 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application;

[0070] Fig.11 A schematic diagram of the computer-readable storage medium structure provided for one embodiment of the present application. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] Example 1

[0073] See also Figure 1 As shown, this embodiment provides a spatial index-based individual loan non-performing asset grid management geographic information retrieval method, including:

[0074] Step S1000, performing initial grid division on the target area to form a first grid set, calculating the non-performing asset density of each grid in the first grid set to obtain a first density distribution; judging whether a first condition for dynamic grid adjustment is triggered according to the first density distribution; if the first condition for dynamic grid adjustment is triggered, performing adaptive re-division on the first grid set to generate a second grid set;

[0075] Furthermore, step S1000 includes:

[0076] Step S1100, obtaining geographical location information of individual loan non-performing assets, performing initial grid division on the target area, and forming a first grid set;

[0077] Furthermore, if Figure 2 As shown, step S1100 includes:

[0078] Step S1110, unifying the geographical location information into longitude and latitude coordinates in the WGS84 coordinate system;

[0079] Step S1120, performing initial grid division on the target area according to the preset first grid granularity parameter and the first grid shape parameter to form n1 grids; n1 is a positive integer;

[0080] Step S1130, extracting core attribute information of each grid; the core attribute information includes grid area, grid quantity, grid number, center coordinates and boundary range;

[0081] Step S1140, counting the number of individual loan non-performing assets falling into each grid, obtaining the ID list of these individual loan non-performing assets, and establishing a grid-asset mapping index;

[0082] Step S1150: forming a first grid set based on the core attribute information and the grid-asset mapping index.

[0083] Specifically, step S1100 first needs to obtain the geographical location information of decentralized personal non-performing loan assets from multiple heterogeneous data sources such as various loan management systems, asset disposal systems, and risk warning systems through database queries, API interface calls, crawlers, etc. Common geographical location representations include longitude and latitude coordinates (such as 116.404, 39.915), structured addresses (such as No. 1, Shuangyushu Dongli, Haidian District, Beijing), unstructured addresses (such as a community near Zhongguancun Square), etc. Since the geographic data collected by different systems may be inconsistent in coordinate systems, formats, etc., a series of coordinate conversion and data cleaning work is required to unify all addresses into longitude and latitude coordinates under the WGS84 geographic coordinate system, eliminate deviations and noise data, and provide unified and standardized input for subsequent grid processing.

[0084] Then, it is necessary to determine the scope of the target area to be divided. It can be flexibly set according to business needs. It can be either nationwide or administrative divisions at different geographical scales such as provinces, cities, districts and counties. At the same time, refer to the spatial distribution characteristics of non-performing assets, the requirements of grid management and other factors to set reasonable first grid granularity parameters and first shape parameters. The first grid granularity parameter controls the geographical size of a single grid. The finer the granularity, the more grids there are, the stronger the spatial expression ability, but the data scale and calculation complexity also increase; the coarser the granularity, the more storage and computing resources can be saved, but the expression ability is lacking, and a trade-off between efficiency and accuracy is needed. Commonly used grid granularity parameters include 1000 meters, 500 meters, 200 meters, 100 meters, etc. The grid shape is generally square or rectangular, which is convenient for slicing and encoding.

[0085] The target area is gridded by using a spatial indexing algorithm to form n1 grids. Common spatial indexing algorithms include quadtree, Hilbert curve, Geohash, etc. These algorithms can map a two-dimensional geographic coordinate space to a one-dimensional linear sequence, which is convenient for grid coding, storage and retrieval. Taking the Geohash algorithm as an example, the grid granularity is set to 6-bit code (about 100 meters x 100 meters). By alternately coding the longitude and latitude binary intervals, a series of unique grid codes (such as wx4g0e) can be generated for the target area. Grids with the same code prefix are also geographically close, so that grids with geographical proximity can be clustered together during storage and query, improving system performance. The quadtree uses a recursive method to divide the square area into four equal parts. Each node represents a geographical area, and the child node represents the sub-area after the parent area is divided into four equal parts. The Hilbert curve fills the two-dimensional plane into a continuous space filling curve, which has a strong geographical aggregation characteristic. These three types of algorithms have their own characteristics in terms of spatial aggregation and query performance. The appropriate spatial indexing method can be selected according to the actual application scenario and data scale.

[0086] After completing the grid division, traverse each generated grid, extract its core attribute information, including grid number, center coordinates, boundary range, etc., and store it in a database or file. At the same time, count the number of individual loan non-performing assets falling into each grid, obtain the ID list of these assets, establish a grid-asset mapping index, and record the distribution of assets contained in the grid. These attribute information and mapping indexes form the first grid set, and subsequent grid retrieval, statistical analysis and other operations will be carried out on this basis.

[0087] Step S1200, calculating the non-performing asset density of each grid in the first grid set to obtain a first density distribution; judging whether to trigger a first condition for dynamic grid adjustment according to the first density distribution;

[0088] Furthermore, if Figure 3 As shown, step S1200 includes:

[0089] Step S1210, traverse each grid in the first grid set, calculate the density of non-performing assets in each grid by the ratio of the number of non-performing assets in the grid to the grid area, generate a first density distribution, and obtain the distribution vector of the first density distribution ; is the non-performing asset density of the ith grid, is the grid number in the first grid set, ;

[0090] Step S1220: Distribute vectors according to the grid area and the first density distribution , calculate the discreteness of the first density distribution;

[0091] Step S1230, comparing the discrete degree of the first density distribution with a preset discrete threshold, if the discrete degree of the first density distribution exceeds the discrete threshold, triggering the first condition of dynamic grid adjustment, otherwise maintaining the current grid division unchanged, and maintaining the first grid set.

[0092] Specifically, after obtaining the first grid set, it is necessary to evaluate whether the existing grid division can well reflect the geographical distribution characteristics of non-performing assets. Among them, asset density is an important evaluation indicator that can quantify the differences and imbalances in asset distribution between different grids. Step S1210 traverses each grid, calculates the number of non-performing assets in the grid in units of grids, divides the number of assets by the grid area, and obtains the non-performing asset density value of each grid. The higher the density value, the more assets are gathered in the unit grid area. Taking the 200-meter grid divided in a certain city-level area as an example, it is found that there are 25 non-performing assets in grid A, 10 in grid B, and 2 in grid C. Their densities are 25 / 0.04=625 / square kilometer, 250 / square kilometer, and 50 / square kilometer, respectively.

[0093] The density values ​​of all grids are aggregated to form the first density distribution, which intuitively shows the differences in the density of asset distribution between different grids. In order to quantitatively evaluate the significance of this difference, it is necessary to further calculate the dispersion index of the density distribution. Commonly used dispersion indexes include variance, standard deviation, etc., which can measure the discreteness or degree of variation of a set of data. Taking variance as an example, the density values ​​of all grids in the first density distribution are averaged, and then the square of the difference between each density value and the mean is calculated, and the average is obtained to obtain the variance of the density distribution. The larger the variance, the higher the dispersion of the density value, and the greater the difference in asset distribution between different grids.

[0094] Finally, the discreteness index is compared with the pre-set threshold. The setting of the threshold needs to comprehensively consider factors such as asset size, grid granularity, and management needs. For example, if the variance is used as an indicator, the threshold can be set to the industry experience value or the previous statistical value. When the current discreteness exceeds the threshold, it means that the existing grid division cannot well reflect the characteristics of asset distribution. There are too many grids with too dense or sparse assets, and adjustments and optimizations are needed. At this time, the first condition for dynamic grid adjustment is triggered, and the next step of the re-division process is entered. When the discreteness meets expectations, it means that the existing grid can reasonably organize the asset distribution and does not need to be adjusted. The current division can be maintained directly. The discrete threshold control adjustment trigger is introduced to strike a balance between algorithm adaptability and system stability.

[0095] The method for calculating the degree of dispersion of the first density distribution comprises:

[0096]

[0097] in:

[0098] D: the degree of dispersion of the first density distribution;

[0099] : The total number of grids in the first grid set;

[0100] : The first grid set The area of ​​the grid;

[0101] : The first grid set The density of non-performing assets in each grid;

[0102] : The weighted average of the non-performing asset density of all grids in the first grid set, where the weight is the proportion of the grid area, and the calculation formula is:

[0103]

[0104] : The first grid set The density weight factor of a grid is calculated as:

[0105]

[0106] when hour, ;when hour, ;

[0107] : The total area of ​​the target area.

[0108] Measure the Density of non-performing assets in each grid Relative to the overall weighted average The greater the deviation, the more unbalanced the grid density distribution is. Introducing log density ratio ,when Significantly higher than hour, Otherwise , making it more sensitive to the effects of density deviation. It is The proportion of the grid area to the whole grid serves as the weight of the denominator and plays a normalization role. The linkage between the grid's own density level and its weight is further strengthened, and the contribution of high-density grids to the overall discreteness is comprehensively amplified.

[0109] When the difference in density between grids is greater and the proportion of high-density grid area is higher, The larger the value, the higher the distribution dispersion. On the contrary, when the difference between the density of each grid is small, or the proportion of high-density grid area is low, The smaller the distribution, the lower the dispersion. The increase of its weight and Both grow exponentially, leading to Grow rapidly.

[0110] Discreteness The influence of grid density difference and area difference is comprehensively considered to quantitatively evaluate the imbalance of the first density distribution. When D exceeds the preset discrete threshold, grid re-division can be triggered to adjust the overly dense or sparse grid and realize automatic response to dynamic changes in density.

[0111] Step S1300: If the first condition for dynamic grid adjustment is triggered, adaptive re-division is performed on the first grid set to generate a second grid set.

[0112] Furthermore, if Figure 4 As shown, step S1300 includes:

[0113] Step S1310, calculating the upper quartile and the lower quartile of the first density distribution, taking the upper quartile as the first density threshold of the first density distribution, and taking the lower quartile as the second density threshold of the first density distribution;

[0114] Step S1320, traversing the first grid set, dividing the grids whose non-performing asset density is higher than the first density threshold into four equal parts, and dividing them into four sub-grids to form a first sub-grid set;

[0115] Step S1330, traversing the first grid set, merging grids whose non-performing asset density is lower than the second density threshold, and merging two grids into one to form a mother grid, thereby obtaining a first mother grid set;

[0116] Step S1340, integrating the first child grid set, the first mother grid set and the grids in the first grid set that have not been adaptively repartitioned to obtain an adjusted grid set, and numbering all grids in the adjusted grid set to form a second grid set.

[0117] The method for numbering all the grids in the adjusted grid set comprises:

[0118] Mark the grids in the first grid set that performs the quartering as the quartered original grids, hide the quartered original grids, and add two subscripts to the numbers of the four newly added subgrids on the basis of the numbers of the quartered original grids to obtain the grid numbers of all the subgrids in the first subgrid set;

[0119] For each mother grid formed by combining two grids into one, the grid number of all mother grids in the first mother grid set is obtained by taking the grid number with the smaller lexicographic order of the original numbers of the two merged grids;

[0120] For the grids in the first grid set that do not perform adaptive re-division, their numbers remain unchanged;

[0121] Summarize the grid numbers of all subgrids in the first subgrid set, the grid numbers of all mother grids in the first mother grid set, and the grid numbers of grids that have not been adaptively repartitioned in the first grid set to obtain the grid numbers of all grids in the adjusted grid set;

[0122] All grids in the adjusted grid set are renumbered according to the lexicographic order of the grid numbers; the lexicographic order of the grid numbers includes an increasing lexicographic order and a decreasing lexicographic order.

[0123] Specifically, the core idea of ​​adaptive gridding is to dynamically adjust the grid division granularity according to the actual distribution of non-performing assets, subdivide the grid in asset-rich areas to improve spatial expression capabilities, and merge the grids in asset-sparse areas to simplify the storage structure, thereby realizing dynamic optimization of spatial index granularity. The upper and lower quartiles of the grid density distribution are calculated and used as the threshold for judging high and low density. Taking the density distribution of 100 grids as an example, they are arranged from small to large according to the density value, and the 25% quantile and 75% quantile density values ​​are taken as thresholds. Grids with less than 25% quantiles belong to low-density grids, those with more than 75% quantiles belong to high-density grids, and those between the two are medium-density grids. The threshold selection needs to dynamically adjust the percentage according to the actual distribution conditions to avoid an imbalance in the proportion of high and low density grid divisions.

[0124] Taking the high-density grid as an example, step S1320 traverses the grid set, and all grids with a non-performing asset density higher than the first threshold (upper quartile) trigger the four-equal division subdivision process. Take a 200-meter grid as an example, assuming that its grid number is 0006, the number of non-performing assets is 40, and the density is 1000 / square kilometer, which is higher than the upper quartile threshold of 800. The grid is divided into four 100-meter * 100-meter sub-grids, and the sub-grid numbers are 000600, 000601, 000610, and 000611. The center coordinates and boundary range of each sub-grid are uniquely determined according to the parent grid and sub-grid number, and then the 40 assets in the parent grid 0006 are traversed, and divided into four sub-grids according to the inclusion relationship between the longitude and latitude and the sub-grid boundary, and the grid asset index is updated, and the original parent grid 0006 is removed from the grid set. After the traversal is completed in this way, the first sub-grid set is formed, and all grids in the set correspond to the areas with the most concentrated assets in the original division.

[0125] For low-density grids below the lower quartile threshold, the opposite "two-in-one" merging strategy is adopted in step S1330. During the traversal, it is found that the density of two adjacent grids 0001 and 0002 is lower than the lower quartile threshold, such as 0001 The density is 10 / square kilometer, and the density of 0002 is 15 / square kilometer, which is lower than the set threshold of 50. At this time, the two grids 0001 and 0002 can be merged into a new mother grid 0001, and the numbering uses the smaller grid number of 0001 and 0002. The center coordinate takes the mean of the two grid centers, the side length takes the sum of the two grid side lengths, and the asset index merges the asset IDs of the two grids 0001 and 0002, thus forming a new grid with doubled granularity. After the merge traversal is completed, the first mother grid set is obtained, which reflects the area with the most sparse assets in the original division. For medium-density grids, since their asset distribution is at a reasonable level, the division and merging process is not triggered for the time being, and the original granularity is maintained unchanged.

[0126] Finally, the first sub-grid set, the first mother grid set, and the unadjusted medium-density grid are summarized and integrated to obtain the second grid set after the granularity is adaptively adjusted. During the integration process, attention is paid to handling some overlapping boundary issues, removing duplicate grids, and renumbering all grids according to the new rules. So far, the whole process of adaptive dynamic adjustment and optimization of the grid set based on the distribution characteristics of non-performing assets has been completed, forming a new generation of grid sets with differentiated granularity and enhanced spatial expression, laying a solid data foundation for subsequent grid management and multidimensional statistical analysis.

[0127] Exemplarily, all grids in the adjusted grid set are numbered:

[0128] Assume that the first grid set contains 8 grids, and the grid numbers and density values ​​sorted by summary density are as follows:

[0129] 0001-5,0002-20,0003-50,0004-88,0005-105,0006-802,0007-810,0008-198;

[0130] After adaptive adjustment, the second grid set is formed:

[0131] High-density grids 0006 and 0007 both exceed the upper quartile threshold of 800, triggering a four-equal subdivision. The original grid is hidden, and the new subgrid is added with a two-digit subscript: 000600, 000601, 000610, 000611; 000700, 000701, 000710, 000711;

[0134] The low-density grids 0001 and 0002 are both below the lower quartile threshold of 50, triggering a two-in-one merge, in which the 0002 grid is hidden and the 0001 grid is retained but the attributes are updated;

[0135] Medium-density grids 0003, 0004, 0005, and 0008 have not reached the adjustment threshold, and their original numbers and attributes remain unchanged;

[0136] Gather all subdivided, merged, and unchanged meshes and renumber them in lexicographic order to form a second mesh set:

[0137] 0001(after merger);

[0138] 0002 (originally 0003);

[0139] 0003 (originally 0004);

[0140] 0004 (originally 0005);

[0141] 0005 (originally 000600);

[0142] 0006 (originally 000601);

[0143] 0007 (originally 000610);

[0144] 0008 (original 000611);

[0145] 0009 (originally 000700);

[0146] 0010 (originally 000701);

[0147] 0011 (originally 000710);

[0148] 0012 (original 000711);

[0149] 0013 (originally 0008).

[0150] In general, the core innovation of this step is to introduce density analysis to evaluate the quality of division, and design a density threshold to automatically trigger grid optimization and adjustment. The algorithm of four-division subdivision and two-in-one coarsening is adopted to achieve dynamic adaptation of grid granularity and overcome the limitations of traditional single static grid. The adjustment strategies of "four-division" and "two-in-one" are clear and concise, with strong interpretability and operability. The combined application of density threshold and quantile can automatically adapt to different asset distribution conditions. The entire algorithm traverses and recursively in grid units, which is easy to implement in parallel and has good computing performance. The adjusted multi-granularity differentiated grid can achieve a more reasonable balance in terms of data scale, retrieval efficiency, statistical granularity, etc., thereby providing strong support for subsequent grid-based fine management.

[0151] Step S2000: construct a first spatial index tree for the first grid set; obtain an incremental grid set according to the first grid set and the second grid set; construct an incremental spatial index tree according to the incremental grid set; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree;

[0152] Furthermore, step S2000 includes:

[0153] Step S2100, constructing a first spatial index tree using the core attribute information of the first grid set and the grid-asset mapping index, wherein each tree node of the first spatial index tree is composed of the core attribute information of the corresponding grid and the ID list of the individual loan non-performing assets;

[0154] Specifically, step S2100 aims to obtain the original spatial index before the grid is adaptively adjusted, providing a basis for subsequent incremental updates. Since the first grid set is generated by static partitioning, its spatial index can be built once the grid partitioning is completed. Common spatial index structures include grid trees, quadtrees, R trees, etc. The core is to achieve hierarchical organization of grid space by establishing a multi-level tree structure, accelerating the retrieval and positioning of grids and assets.

[0155] Taking the quadtree index as an example, starting from the root node, each parent node has four child nodes, which represent the four sub-regions formed by equally dividing the rectangular area corresponding to the parent node in the horizontal and vertical directions. The grid with the coarsest granularity and the highest level in the first grid set is used as the root node of the quadtree. Each node contains the core attribute information such as the number, center coordinates, and side length of the grid, as well as the list of personal loan non-performing asset IDs directly contained in the grid. Then traverse the grids of the next level. For the grid with the second coarsest granularity in the first grid set, if it can be completely contained in a parent grid in space, it will be used as a child node of the parent grid, and the parent-child topological relationship between the two grids will be recorded. The recursive traversal is repeated until the index of the finest granularity grid is completed, forming a complete multi-layer quadtree index structure.

[0156] By pre-building a static grid index, the subsequent retrieval complexity of grids and assets is reduced to O(logn) level (logarithmic), which is significantly better than O(n) sequential traversal search (linear). At the same time, the index records the complete grid space topology, which facilitates advanced spatial queries such as neighborhood analysis and buffer analysis.

[0157] Step S2200, obtaining an incremental grid set according to the first grid set and the second grid set;

[0158] Furthermore, if Figure 5 As shown, step S2200 includes:

[0159] Step S2210, traversing the second grid set, extracting grids whose numbers do not exist in the first grid set, to form a newly added grid set;

[0160] Step S2220, traversing the first grid set, extracting grids whose numbers do not exist in the second grid set, to form a deleted grid set;

[0161] Step S2230, traversing the second grid set, extracting the grids whose numbers are the same as those of a grid in the first grid set but whose core attribute information has changed, to form an updated grid set;

[0162] Step S2240, performing a union operation on the newly added grid set, the deleted grid set, and the updated grid set to obtain an incremental grid set.

[0163] Exemplarily, the example in step S1300 is continued:

[0164] Step S2210, extracting a newly added grid set:

[0165] Traverse the second grid set and extract the grids numbered 0005-0012, because they come from the four equal parts of the high-density grids 0006 and 0007 and do not exist in the first grid set:

[0166] Added grid set = {0005, 0006, 0007, 0008, 0009, 0010, 0011, 0012};

[0167] Step S2220, extract and delete the grid set:

[0168] Traverse the first grid set and extract the three grids 0002, 0006, and 0007, because they have been merged or subdivided in the second grid set and no longer exist:

[0169] deletegridset={0002,0006,0007};

[0170] Step S2230, extract and update the grid set:

[0171] Traverse the second grid set and extract the grid numbered 0001. Although the grid number is the same as the grid number 0001 in the first grid set, its area, number of assets, density and other attributes have changed due to the absorption and merger of grid number 0002, so it is an updated grid:

[0172] updategridset = {0001};

[0173] Step S2240, merging and summarizing the incremental grids:

[0174] Take the union of the three types of grid number sets, namely, added, deleted, and updated, and get the complete incremental grid set:

[0175] Incremental grid set = new grid set ∪ delete grid set ∪ update grid set

[0176] ={0001,0002,0005,0006,0007,0008,0009,0010,0011,0012}.

[0177] This set covers all grids that have changed (added / deleted / updated) between the first and second grid sets, and is the focus of subsequent analysis. As for the three unchanged grids numbered 0003, 0004, and 0013, since their numbers and properties remain unchanged, the old grid index can be directly used to avoid repeated processing.

[0178] Specifically, the adaptively adjusted second grid set may have three types of changes compared to the original first grid set: grid addition, deletion, and update. The newly added grids come from the four-equal division of the high-density area, which forms a new numbered sub-grid after the division. The deleted grids come from the two-in-one merger of the low-density area, and the original coarse-grained grids no longer exist after the merger. The updated grids correspond to the medium-density grids that have not triggered the adjustment. The numbers remain unchanged, but the attribute values ​​such as asset density may change.

[0179] The identification of incremental grid sets is essentially to extract the difference between the first and second grid sets by comparing the grid numbers. You can first use a hash set to store all the grid numbers of the two grid sets, and then perform a hash difference operation to obtain the newly added number set that only exists in the second grid set, and the deleted number set that only exists in the first grid set. Then traverse the two grid sets, extract the grids with the same corresponding numbers but inconsistent attribute values, and generate an updated number set. Finally, merge these three types of number sets to obtain a complete incremental grid set, which covers all the grids that have changed.

[0180] The time complexity of this incremental recognition method is O(m+n), where m and n are the number of grids in the two grid sets. Efficient difference comparison is achieved through hash and union operations, avoiding the O(mn) overhead of pairwise cross comparison. By focusing on the changed grids, the processing objects of subsequent index reconstruction are greatly reduced, the resource consumption of index updates is reduced, and the response speed is improved. At the same time, the three types of changed grids are uniformly organized into incremental sets, which facilitates the use of a consistent algorithm framework for processing and reduces program complexity.

[0181] Step S2300, constructing an incremental spatial index tree according to the incremental grid set;

[0182] Furthermore, if Figure 6 As shown, step S2300 includes:

[0183] Step S2310, traversing each grid in the incremental grid set, extracting the second core attribute information of each grid, and constructing a second grid-asset mapping index;

[0184] Step S2320, constructing an incremental spatial index tree according to the second core attribute information of the incremental grid set and the second grid-asset mapping index;

[0185] Step S2330, traverse and serialize the incremental spatial index tree to generate an incremental index data file that is easy to store and transmit.

[0186] Specifically, the construction of the incremental spatial index is essentially the same as the construction method of the first spatial index tree, the difference is that the index object is reduced from the first grid set to the incremental grid set. Taking the quadtree as an example, each grid in the incremental grid set is traversed to extract its core attribute information such as number, coordinates, boundaries, etc. According to the grid numbering rules and spatial topological relationships, its parent-child hierarchical association in the quadtree is determined, and a multi-layer tree index is gradually generated.

[0187] Considering the orientation and adjacency relationship between the new and old grid sets, the root of the incremental quadtree should be the grid with the largest area in the incremental grid set. Each layer of sub-nodes corresponds to a sub-grid whose length and width are halved successively, that is, the side length of the k-th layer node is 1 / 2 of the side length of the root. k In particular, for newly added grids in the grid set, it is necessary to regard it as the root of the tree to generate a new subtree index; for deleted grids and updated grids in the grid set, the corresponding original subtree index is pruned and modified to complete the dynamic adjustment of the subtree. This iteration is repeated until the last leaf node, and an incremental spatial index tree reflecting the incremental changes can be obtained.

[0188] During the index tree construction process, attention should be paid to handling the connection problems of grid boundaries, such as the overlap and intersection of new and old grids at the common boundaries. If necessary, boundary tolerance should be introduced to absorb and merge grids to avoid the appearance of narrow and long fragmented grids. In addition, the incremental index tree should be as consistent as possible with the original index tree (the first spatial index tree) in structure, and the hierarchical relationship of the original tree should be used in the common area to achieve a smooth logical transition between the new and old indexes, which is convenient for the matching and fusion of tree structures in subsequent index merging operations.

[0189] Finally, the incremental index tree is traversed and serialized, and a linearized tree node sequence is generated through the breadth-first or depth-first principle. According to the grid attribute fields of the nodes, it is serialized into a data file that is easy to store and transmit. Common serialization methods include JSON, XML, binary objects, etc., which need to balance the data volume and parsing efficiency. The serialized incremental index file can be stored independently as a patch file of the original index, or it can be directly merged with the original index. The two modes can be selected according to the frequency and real-time requirements of the full index update in actual applications.

[0190] Step S2400: merge the first spatial index tree and the incremental spatial index tree to form a second spatial index tree.

[0191] Specifically, the merging of the original index (first spatial index) and the incremental index is a combination of node matching, attribute coverage, subtree splicing and other operations based on tree topology, which can gradually realize the combing and synchronization of grid changes in the index. Traverse the original index tree, and for updated grid nodes in the incremental index tree, replace the corresponding core attribute values ​​in the original index tree (first spatial index tree) with the core attribute values ​​in the incremental index tree to complete the index merging of the updated grid; the core attributes include the center coordinates, geocoding, statistical indicators, etc. of the grid. Traverse the original index tree, and for the grid nodes deleted in the incremental index tree, remove the corresponding subtree branches in the original index tree, and repair the tree structure upward to complete the index merging of the deleted grid; for the deleted grid, the corresponding tree branches need to be removed from the index tree, and the grid attributes of the parent node and ancestor node of the deleted node are recursively modified. If the deletion of a child node makes the parent node no longer meet the quadtree splitting condition, the four subtrees of the parent node should be merged and simplified to avoid the index tree from degenerating into a "burr shape". The newly added grid nodes in the incremental index tree, together with their subordinate subtree branches, are inserted into the corresponding positions of the original index tree to complete the index merging of the newly added grids; for the newly added grids, since there is no corresponding node in the original tree, the subtree representing the newly added grid needs to be inserted into the specified position of the original tree as a whole. The selection of the insertion position needs to consider the azimuthal adjacency relationship between the new and old grids, and try to make the newly added nodes close to the adjacent parent nodes to reduce the local imbalance of the index tree.

[0192] When deleting and inserting nodes, it is easy to cause an unbalanced state in the local index tree, such as a side chain of the tree is too long, some intermediate nodes have too many child nodes, etc. At this time, the index tree needs to be adaptively optimized. After the node insertion and deletion operations are completed, the merged index tree is traversed, and the number of child nodes of each node is checked layer by layer from the bottom to the top. For nodes whose number of child nodes exceeds the threshold, a mixed index is introduced to achieve adaptive optimization of the index tree. The core of the optimization is to identify the number of child nodes of each node during the bottom-up traversal of the index tree, and trigger the optimization process when the number of child nodes exceeds the set threshold. For nodes with too many child nodes, a mixed index is introduced. On the basis of the quadtree index, a subtree index is further nested for the child node. Commonly used ones are MX-quadtree, R-tree, etc. Within the sub-index, the same idea can be continued to be applied for multi-level recursive subdivision to achieve further index optimization of local high-density grids.

[0193] Through the dynamic reflection of grid updates, deletions, and additions in the index tree, as well as the adaptive optimization of the index tree, a complete second spatial index tree is finally formed. This index tree encompasses all grids in the second grid set, integrates the spatial topological relationship of the original grid set, absorbs the dynamic information of incremental changes, and can continuously maintain efficient index support for the latest grid status, laying a solid and efficient data foundation for various complex spatial queries and statistical analysis.

[0194] For example, in a certain adaptive gridding adjustment, after the splitting and merging of the grids, it was identified that 5% of the grids had changed, forming an incremental grid set consisting of 500 grids. The attribute extraction and mapping index construction of these 500 grids can be completed within 1 second, and the corresponding incremental index tree can be constructed within 2 seconds based on the quadtree algorithm. By serializing the incremental tree, a compressed index update file of about 5MB is generated. Read the last full index tree, traverse its nearly 2000 tree nodes, replace the attributes of 50 updated nodes, and remove the subtree branches corresponding to 20 deleted nodes. At the same time, traverse the incremental index tree in turn, and dynamically insert its 400 newly added nodes into the corresponding positions of the original index tree. During the insertion process, it was found that the number of child nodes of 20 nodes exceeded the preset threshold of 100, that is, the grid was too dense locally. At this time, the R-tree index was automatically introduced for these nodes for optimization, and each R-tree was nested with a 10x10 quadgrid index. So far, the seamless merging of full and incremental index trees has been achieved, and the original index tree has been expanded to 2,300 nodes and 8 layers deep. The adaptive grid index construction of 50,000 irregular grids in an urban area of ​​10,000 square kilometers has been completed. The whole process takes only 5 seconds, and the grid retrieval efficiency is 150 times higher than that of the sequential traversal method, and the index tree occupies less than 30MB of storage space.

[0195] Step S3000, receiving a geographic information retrieval request, performing a retrieval according to the second spatial index tree, and generating a retrieval result; judging whether the second condition for dynamic adjustment of the grid is triggered according to the retrieval result; if the second condition for dynamic adjustment of the grid is triggered, dynamically adjusting the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-dividing the target area into initial grids; if the second condition for dynamic adjustment of the grid is not triggered, completing this retrieval and outputting the retrieval result.

[0196] Furthermore, step S3000 includes:

[0197] Step S3100, receiving a geographic information search request, wherein the geographic information search request includes the geographic location of the target individual loan non-performing assets, a target search range parameter, and a target grid granularity parameter;

[0198] Step S3200, extracting a target grid granularity parameter as a second grid granularity parameter, comparing the second grid granularity parameter with the first grid granularity parameter, and determining whether the second grid granularity parameter is consistent with the first grid granularity parameter;

[0199] Step S3300: if the second grid granularity parameter is consistent with the first grid granularity parameter, directly perform the search using the second spatial index;

[0200] Step S3400: if the second grid granularity parameter is inconsistent with the first grid granularity parameter, the second grid set is re-divided according to the second grid granularity parameter to form a third grid set, and a third spatial index tree corresponding to the third grid set is established, and the third spatial index tree is used to perform retrieval.

[0201] Specifically, the geographic location of the target individual loan non-performing assets is used to locate the search starting point, the target search range parameter is used to determine the search area boundary, and the target grid granularity parameter is used to specify the expected search grid size; an adaptive index granularity adjustment strategy is introduced to dynamically match the optimal index granularity according to the granularity requirements of different search requests, and achieve a better balance between query efficiency and storage overhead. The core idea is to compare the target grid granularity parameter of the search request with the first grid granularity parameter of the existing index to determine whether the current index granularity can meet the current search requirements. If the two granularity parameters have the same value, it means that the grid granularity of the existing second spatial index just matches the request. At this time, the index can be used directly to perform the search, avoiding the overhead of repeated indexing. On the contrary, if the target granularity is inconsistent with the current index granularity, it is necessary to dynamically adjust the grid division scheme, subdivide or merge the existing grids, generate a third grid set that matches the target granularity, and build the corresponding third spatial index, so as to ensure that the search granularity matches the request requirements. This adaptive adjustment can significantly improve the adaptability and flexibility of the index.

[0202] Taking a business system as an example, a city was gridded using the first grid granularity parameter of 500 meters in the early stage, forming a second grid set containing 2,000 grids. When a search request with a target granularity of 500 meters is received, the parameter is extracted as the second grid granularity parameter. If it is found to be consistent with the first grid granularity parameter, the existing second spatial index is directly used to complete the search. When the request granularity becomes 200 meters, since it does not match the current index granularity, the second grid set needs to be further divided into four equal parts, and refined into a third grid set of 8,000 200-meter grids, and then a new third spatial index is built for it, so as to match the target granularity to complete the search. This adaptive multi-granularity, multi-level index structure can dynamically balance the search accuracy and search efficiency.

[0203] The method for determining whether the second grid granularity parameter is consistent with the first grid granularity parameter comprises:

[0204] Assume the first grid size parameter is , the second grid size parameter is , introduce two auxiliary variables and , Indicates the change range of grid size, Indicates the direction of change of grid granularity;

[0205]

[0206] in, is a symbolic function, when hour, Pick hour, Pick When 0.

[0207] Define the grid size consistency judgment function :

[0208]

[0209] in, ,when When and — to; when When and Inconsistent; when When and It is in a critical consistency state, which means that in this state, the particle size parameters of the two are "marginally consistent", which can be understood as close to consistency or not yet completely consistent.

[0210] , characterization Relative to The degree of change; and The greater the difference, The bigger. and The smaller value in is normalized to make the denominator and and The absolute size of the change has nothing to do with it, but only with the relative magnitude of the change.

[0211] , characterization Relative to Is it getting bigger, the same, or getting smaller.

[0212] The value range is ,along with As it increases from 0 to 2, the term decreases monotonically. and When equal, ;when yes When it is twice or half of ;when yes When it is three times or one third of , This term describes the negative impact of the change in grid size on consistency.

[0213] The value range is when When , this item follows Increases with increase, which has a positive effect on consistency; when When , this item follows increases and decreases, further reducing the consistency; when When , which is always equal to 1 and does not affect consistency. This enables the judgment formula to distinguish between the two situations of increasing and decreasing grid granularity.

[0214] The -1 in the above formula is the translation factor. The zero point of is shifted to -1, so that When expressing agreement, It indicates inconsistency, which is convenient for judgment.

[0215] In general, the grid size changes The larger the change direction coefficient The further away from 0, the consistency judgment function The smaller, and The lower the consistency. hour, , indicating complete agreement.

[0216] This formula cleverly introduces trigonometric functions and exponential functions, quantitatively describes the nonlinear influence of the amplitude and direction of the change in the grid granularity parameters on the consistency judgment, and can more comprehensively and meticulously evaluate the difference between the first grid granularity and the second grid granularity, providing an important mathematical criterion for adaptive gridding. At the same time, the judgment formula is easy to calculate, and the required parameters are easy to directly extract from the grid division process, which has strong practical value.

[0217] Step S3500, performing a search and generating search results;

[0218] Furthermore, if Figure 7 As shown, step S3500 includes:

[0219] Step S3510, determining the spatial boundary of a rectangular search area centered on the geographical location of the target non-performing asset according to the target search range parameters;

[0220] Step S3520, calculating the grid numbers of the starting grid and the ending grid covered by the rectangular search area, and using them as the search starting grid number and the search ending grid number respectively;

[0221] Step S3530, using the second spatial index tree or the third spatial index tree, traverse the grid set covered by the search start grid number to the search end grid number to form a candidate grid set;

[0222] Step S3540, extracting the geographical information of non-performing assets contained in the candidate grid set and generating search results.

[0223] Specifically, step S3500 makes full use of the grid coding and geographic spatial mapping characteristics of the spatial index to achieve rapid positioning and filtering for any search range. By parsing the target search range parameters of the request, such as the center point coordinates and the search radius, the geographic area covered by the range is calculated. The minimum circumscribed rectangle of the area is further located to obtain the geographic boundary coordinates of the rectangle. By reversing the grid coding through the coordinates, the start and end numbers of all grids covered by the rectangle can be determined. This number mapping avoids complex spatial operations between geographic areas and grids.

[0224] The spatial index uses methods such as Geohash coding to achieve one-dimensional mapping between grid numbers and geographic space, which is convenient for rapid retrieval. Assuming that the target non-performing asset geographic location of the retrieval request is (116.500, 39.900), the retrieval radius is 3000 meters, the spatial boundary of the rectangular retrieval area can be obtained through coordinate calculation as [(116.485, 39.890), (116.515, 39.910)], and the corresponding Geohash coding boundary is [wx4g2, wx4g8]. Then wx4g2 is the starting grid number of the retrieval, and wx4g8 is the ending number. Next, all grid numbers within wx4g2~wx4g8 are traversed, and the candidate grids corresponding to these numbers are retrieved through the second or third spatial index, and all the non-performing asset geographic information contained in the grid is summarized, that is, the accurate asset distribution in the target area of ​​this retrieval is obtained, and efficient retrieval is completed. This index retrieval path is simple and avoids complex geometric operations on geographic areas and grid sets.

[0225] Step S3600, based on the search results, determine whether the second condition for dynamic grid adjustment is triggered; if the second condition for dynamic grid adjustment is triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the search quality evaluation parameter, and re-perform the initial grid division on the target area; if the second condition for dynamic grid adjustment is not triggered, complete this search and output the search results.

[0226] Further, step S3600 includes:

[0227] Step S3610, cluster analysis is performed on the search results to generate search quality evaluation parameters that reflect the rationality of grid division within the time and space range; the search quality evaluation parameters include cluster grid density dispersion coefficient GDDC, cluster asset density dispersion coefficient ADDC, intra-cluster grid granularity variation coefficient GGVC, and inter-cluster grid granularity variation coefficient GGVB;

[0228] Step S3620, comparing the retrieval quality evaluation parameter with a preset adjustment threshold to determine whether the second condition for dynamic grid adjustment is triggered.

[0229] Furthermore, if Figure 8 As shown, step S3620 includes:

[0230] Step S3621, setting a grid density discrete threshold for the clustering grid density discrete coefficient GDDC ;

[0231] Step S3622: Set the asset density discrete threshold for the cluster asset density discrete coefficient ADDC ;

[0232] Step S3623, setting the intra-cluster grid size variation threshold for the intra-cluster grid size variation coefficient GGVC ;

[0233] Step S3624, setting the inter-cluster grid size variation coefficient GGVB for the inter-cluster grid size variation coefficient GGVB ;

[0234] Step S3625: When two or more retrieval quality evaluation parameters exceed their respective thresholds, a second condition for triggering dynamic grid adjustment is determined.

[0235] Specifically, step S3600 introduces a closed-loop mechanism for grid division and index optimization for search results. By statistically analyzing the grid and asset distribution within the spatial and temporal range covered by this search, evaluating the adaptability of the existing grid granularity to the region, and dynamically triggering the adjustment of the grid division scheme and index, a fine-grained local adaptive optimization is achieved. The core is to evaluate the index performance based on actual search feedback, extract search quality evaluation parameters, and determine whether the grid and index structure need to be adjusted through threshold comparison.

[0236] Perform cluster analysis on the search results. Common clustering algorithms include K-Means and DBSCAN. The search results are automatically divided into several clusters based on the distance between grids and assets in geographic space as a similarity measure. The clustering results can reveal the differences between different sub-regions within the search area.

[0237] Taking the query of a certain city-level area as an example, a total of 120 grids and 2,000 non-performing assets were retrieved. After clustering using the K-Means algorithm, it was found that they can be divided into three clusters, corresponding to the central urban area, suburban area and remote suburbs. Further analysis found that the central urban area cluster contains 40 grids, the grid granularity is mostly 100 meters, and there are 30 assets per grid on average; the suburban area cluster contains 50 grids, the grid granularity ranges from 200 to 500 meters, and there are 20 assets per grid on average; the remote suburban area cluster contains 30 grids, the grid granularity is mostly 1,000 meters, and there are 5 assets per grid on average. There are obvious differences in the number of grids, granularity and asset density among different clusters.

[0238] Retrieval quality evaluation parameters are a set of quantitative indicators used to measure the aggregation characteristics and discreteness of retrieval results in the time and space dimensions, and indirectly reflect the adaptability of the current grid division to the distribution of non-performing assets. Retrieval quality evaluation parameters include:

[0239] 1) Grid Density Dispersion Coefficient of Clusters (GDDC): Perform spatiotemporal clustering on the search results, calculate the ratio of the standard deviation of the grid density to the mean in each cluster, and then take the weighted average of the GDDC of all clusters, where the weight is the ratio of the number of grids contained in the cluster to the total number of grids. The larger the GDDC, the greater the difference in grid density within different clusters, and the lower the rationality of the grid division.

[0240] Taking the above urban cluster as an example, there are 40 grids in total, with an average of 16 grids per square kilometer, but the grid density of each grid varies greatly, with the highest density reaching 25 per square kilometer and the lowest being only 5. The calculated GDDC is 35%, reflecting that the granularity of the grid division within the central urban area is not balanced.

[0241] 2) Asset Density Dispersion Coefficient of Clusters (ADDC): Similar to GDDC, the ratio of the standard deviation of asset density within each cluster to the mean is calculated, and then the weighted average is taken. The larger the ADDC, the greater the difference in asset density within different clusters, and the less accurate the grid division is in describing asset distribution.

[0242] Taking urban clustering as an example, there are an average of 480 non-performing assets per square kilometer, but the asset density of each grid varies greatly, with the highest reaching 800 and the lowest being only 100. The calculated ADDC is 55%, reflecting that the distribution of assets in the urban area is very unbalanced, and there may be overly dense and sparse areas.

[0243] 3) Grid Granularity Variation Coefficient within Clusters (GGVC): Calculate the ratio of the standard deviation of the grid granularity within each cluster to the mean, and then take the weighted average. The larger the GGVC, the more grids with large granularity differences are mixed in the same cluster, and the worse the consistency of the division.

[0244] Taking the suburban cluster as an example, there are 50 grids with granularity ranging from 100 meters to 1000 meters and an average granularity of 400 meters. The calculated GGVC is as high as 45%, indicating that the grid division standards of each sub-area within the suburban area are not consistent, some are too detailed, and some are too rough.

[0245] 4) Grid Granularity Variation Coefficient between Clusters (GGVB): Calculate the ratio of the standard deviation of the grid granularity means of different clusters to the overall mean. The larger the GGVB, the greater the difference in grid granularity between different clusters, and the worse the uniformity of the division.

[0246] Taking the above three clusters as an example, the average grid granularity of the central urban area, suburban area and remote suburban area is 100 m, 400 m and 1000 m respectively, the overall average granularity is 500 m, and the calculated GGVB is 80%, reflecting that the grid division scales of the three regions are quite different and lack uniformity.

[0247] These evaluation parameters reflect the degree of fit between grid division and NPL distribution from different perspectives. The dispersion coefficient eliminates the influence of dimension and value by calculating the ratio of standard deviation to mean, so that the results under different time and space ranges and different data scales can be compared horizontally. The coefficient of variation further considers the influence of the dispersion degree within the indicator on the overall.

[0248] After obtaining the retrieval quality assessment parameters, it is necessary to compare them with the preset adjustment threshold to determine whether to trigger the second condition of dynamic grid adjustment. The setting of the adjustment threshold needs to comprehensively consider factors such as the scale of non-performing assets, the speed of distribution change, and the characteristics of spatiotemporal clustering. It can be estimated and calibrated based on historical data and business experience. For example, thresholds can be set for GDDC, ADDC, GGVC, and GGVB respectively. When two or more parameters exceed their respective thresholds, it is considered that the current grid division and indexing are no longer suitable for asset distribution and need to be dynamically adjusted and optimized.

[0249] If it is confirmed that the dynamic adjustment of the grid is triggered, it is necessary to further analyze the values ​​of the evaluation parameters, locate the cause of the problem, and dynamically adjust the parameters of the grid division accordingly, such as the first grid granularity, the first grid shape, etc. The adjustment direction of the parameters is inversely related to the evaluation results, that is, the larger the GDDC and ADDC, the grid granularity should be appropriately reduced to increase the coverage of a single grid; the larger the GGVC and GGVB, the grid granularity should be appropriately increased to refine the division granularity of different areas; horizontally compare the relative sizes of GDDC and ADDC. If GDDC is larger, it means that the asset distribution within the grid is more unbalanced, and the grid granularity should be reduced first; if ADDC is larger, it means that the asset distribution between grids is more unbalanced, and the grid shape should be adjusted first to adapt to the spatial distribution direction of the assets.

[0250] The adjustment range can be determined by referring to the extent to which the evaluation parameters exceed the threshold and historical adjustment experience. Too large an adjustment range may cause oscillation and frequent reconstruction of the grid division, while too small an adjustment range will make it difficult to keep up with the pace of changes in asset distribution. It is necessary to seek a dynamic balance between efficiency and effect.

[0251] After the parameter adjustment is completed, return to step S1000, re-dynamically mesh based on the new parameters, and update the spatial index accordingly. Through the dynamic closed loop of meshing, index construction, retrieval evaluation, parameter adjustment and other steps, the continuous adaptation and optimization of the grid and asset distribution can be achieved, and the timeliness and accuracy of retrieval and management can be continuously improved. Even if the asset distribution pattern changes, the grid and index can be adaptively corrected through the dynamic adjustment mechanism to ensure that the quality of the retrieval results is stable and controllable.

[0252] Through the adaptive retrieval optimization in step S3000, the method can achieve dynamic adaptation of grid division, spatial indexing and actual retrieval mode at multiple granularity levels and multiple spatiotemporal scales, and can continue to maintain efficient and accurate retrieval performance when asset distribution and retrieval requirements change dramatically. This full-process dynamic closed-loop design enables the grid management method to have the ability of self-evolution and self-optimization, providing highly adaptable geographic information retrieval support for the precise management of non-performing assets.

[0253] Exemplary:

[0254] For a county with increasing business monitoring requirements and frequent changes in asset status. Initially, a coarse-grained grid of 1,000 meters was used to organize and manage about 5,000 non-performing assets in the entire county. After one month of operation, it was found that the assets in some grids in the frequently searched areas of hot towns were highly dense and extremely unevenly distributed, and the asset density dispersion coefficient was as high as 0.85, which has become a bottleneck restricting the search performance. Therefore, the grid granularity of these towns was adaptively adjusted to 200 meters, and the top 10 high-density grids were further subdivided to form a multi-level quadtree index. After the update, it was searched again and it was found that the original dozens of abnormally high-density grids were divided into hundreds of fine grids, and the dispersion coefficient dropped to below 0.3, and the spatial expression ability of the over-dense area was greatly improved. At the same time, the dense index at the township level accelerated 70% of the hotspot access. The areas that did not trigger the adjustment maintained a sparse grid of 1,000 meters, saving storage resources.

[0255] For comprehensive statistical analysis tasks across the entire county, coarse-grained indexes can still be used for rapid response. An asset inventory task requires the distribution calculation coverage of all assets in the county. The entire domain is searched with a 1,000-meter index first, and the asset distribution profile at the township level can be obtained in seconds without dynamic adjustment. For inventory in local areas, fine-grained index reconstruction is triggered in real time to achieve dynamic refinement at the minute level. Finally, supported by a multi-granular, adaptive dynamic grid index system, the county has achieved accurate and efficient geographic information services for the entire process of non-performing asset management, and the retrieval performance has been improved several times compared to traditional solutions.

[0256] This case fully demonstrates the adaptive ability of this method to cope with complex and changing business scenarios. This dynamic optimization mechanism gives the grid management of non-performing assets stronger environmental adaptability and better spatiotemporal performance. It is a highly robust and highly available geographic information intelligent organization method, which will play an important role in accurate asset supervision, efficient risk response, and scientific decision-making assistance.

[0257] Example 2

[0258] This embodiment provides a grid management geographic information retrieval system for individual loan non-performing assets based on spatial indexing on the basis of embodiment 1. Fig. 9 As shown, including:

[0259] Grid division module: used to perform initial grid division on the target area to form a first grid set, calculate the non-performing asset density of each grid in the first grid set, and obtain a first density distribution; according to the first density distribution, determine whether the first condition for dynamic grid adjustment is triggered; if triggered, perform adaptive re-division on the first grid set to generate a second grid set; if not triggered, keep the first grid set;

[0260] Spatial index module: used to construct a first spatial index tree for a first grid set; obtain an incremental grid set according to the first grid set and the second grid set; construct an incremental spatial index tree according to the incremental grid set; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree;

[0261] Retrieval module: used to receive geographic information retrieval requests, perform retrieval according to the second spatial index tree, and generate retrieval results; based on the retrieval results, determine whether the second condition for dynamic grid adjustment is triggered; if triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-initialize the grid division of the target area; if not triggered, complete this retrieval and output the retrieval results.

[0262] In the grid division module, the initial grid division of the target area to form a first grid set includes:

[0263] Step S1110, obtaining the geographical location information of the individual loan non-performing assets, and unifying the geographical location information into longitude and latitude coordinates in the WGS84 coordinate system;

[0264] Step S1120, performing initial grid division on the target area according to the preset first grid granularity parameter and the first grid shape parameter to form n1 grids; n1 is a positive integer;

[0265] Step S1130, extracting core attribute information of each grid; the core attribute information includes grid area, grid quantity, grid number, center coordinates and boundary range;

[0266] Step S1140, counting the number of individual loan non-performing assets falling into each grid, obtaining the ID list of these individual loan non-performing assets, and establishing a grid-asset mapping index;

[0267] Step S1150: forming a first grid set based on the core attribute information and the grid-asset mapping index.

[0268] In the grid division module, the non-performing asset density of each grid in the first grid set is calculated to obtain a first density distribution; and according to the first density distribution, the first condition for determining whether to trigger dynamic grid adjustment includes:

[0269] Step S1210, traverse each grid in the first grid set, calculate the density of non-performing assets in each grid by the ratio of the number of non-performing assets in the grid to the grid area, generate a first density distribution, and obtain the distribution vector of the first density distribution ; is the non-performing asset density of the ith grid, is the grid number in the first grid set, ;

[0270] Step S1220: Distribute vectors according to the grid area and the first density distribution , calculate the discreteness of the first density distribution;

[0271] Step S1230, comparing the discrete degree of the first density distribution with a preset discrete threshold, if the discrete degree of the first density distribution exceeds the discrete threshold, triggering the first condition of dynamic grid adjustment, otherwise maintaining the current grid division unchanged, and maintaining the first grid set.

[0272] In the grid division module, the step of performing adaptive re-division on the first grid set to generate the second grid set includes:

[0273] Step S1310, calculating the upper quartile and the lower quartile of the first density distribution, taking the upper quartile as the first density threshold of the first density distribution, and taking the lower quartile as the second density threshold of the first density distribution;

[0274] Step S1320, traversing the first grid set, dividing the grids whose non-performing asset density is higher than the first density threshold into four equal parts, and dividing them into four sub-grids to form a first sub-grid set;

[0275] Step S1330, traversing the first grid set, merging grids whose non-performing asset density is lower than the second density threshold, and merging two grids into one to form a mother grid, thereby obtaining a first mother grid set;

[0276] Step S1340, integrating the first child grid set, the first mother grid set and the grids in the first grid set that have not been adaptively repartitioned to obtain an adjusted grid set, and numbering all grids in the adjusted grid set to form a second grid set.

[0277] The spatial index module includes: a first spatial index unit;

[0278] The first spatial index unit is used to construct a first spatial index tree using the core attribute information of the first grid set and the grid-asset mapping index, wherein each tree node of the first spatial index tree consists of the core attribute information of the corresponding grid and the ID list of individual loan non-performing assets.

[0279] In the spatial index module, obtaining the incremental grid set according to the first grid set and the second grid set includes:

[0280] Step S2210, traversing the second grid set, extracting grids whose numbers do not exist in the first grid set, to form a newly added grid set;

[0281] Step S2220, traversing the first grid set, extracting grids whose numbers do not exist in the second grid set, to form a deleted grid set;

[0282] Step S2230, traversing the second grid set, extracting the grids whose numbers are the same as those of a grid in the first grid set but whose core attribute information has changed, to form an updated grid set;

[0283] Step S2240, performing a union operation on the newly added grid set, the deleted grid set, and the updated grid set to obtain an incremental grid set.

[0284] In the spatial index module, constructing an incremental spatial index tree according to the incremental grid set includes:

[0285] Step S2310, traversing each grid in the incremental grid set, extracting the second core attribute information of each grid, and constructing a second grid-asset mapping index;

[0286] Step S2320, constructing an incremental spatial index tree according to the second core attribute information of the incremental grid set and the second grid-asset mapping index;

[0287] Step S2330, traverse and serialize the incremental spatial index tree to generate an incremental index data file that is easy to store and transmit.

[0288] In the search module, the geographic information search request includes the geographic location of the target individual loan non-performing assets, the target search range parameter and the target grid granularity parameter;

[0289] The performing of the search and generating the search results comprises:

[0290] Step S3100, receiving a geographic information search request, wherein the geographic information search request includes the geographic location of the target individual loan non-performing assets, a target search range parameter, and a target grid granularity parameter;

[0291] Step S3200, extracting a target grid granularity parameter as a second grid granularity parameter, comparing the second grid granularity parameter with the first grid granularity parameter, and determining whether the second grid granularity parameter is consistent with the first grid granularity parameter;

[0292] Step S3300: if the second grid granularity parameter is consistent with the first grid granularity parameter, directly perform the search using the second spatial index;

[0293] Step S3400, if the second grid granularity parameter is inconsistent with the first grid granularity parameter, then re-dividing the second grid set according to the second grid granularity parameter to form a third grid set, and establishing a third spatial index tree corresponding to the third grid set, and using the third spatial index tree to perform retrieval;

[0294] Step S3510, determining the spatial boundary of a rectangular search area centered on the geographical location of the target non-performing asset according to the target search range parameters;

[0295] Step S3520, calculating the grid numbers of the starting grid and the ending grid covered by the rectangular search area, and using them as the search starting grid number and the search ending grid number respectively;

[0296] Step S3530, using the second spatial index tree or the third spatial index tree, traverse the grid set covered by the search start grid number to the search end grid number to form a candidate grid set;

[0297] Step S3540, extracting the geographical information of non-performing assets contained in the candidate grid set and generating search results.

[0298] In the retrieval module, the second condition for determining whether to trigger dynamic grid adjustment according to the retrieval results includes:

[0299] Step S3610, cluster analysis is performed on the search results to generate search quality evaluation parameters that reflect the rationality of grid division within the time and space range; the search quality evaluation parameters include cluster grid density dispersion coefficient GDDC, cluster asset density dispersion coefficient ADDC, intra-cluster grid granularity variation coefficient GGVC, and inter-cluster grid granularity variation coefficient GGVB;

[0300] Step S3620, comparing the retrieval quality evaluation parameter with a preset adjustment threshold to determine whether the second condition for dynamic grid adjustment is triggered.

[0301] The step S3620 includes:

[0302] Step S3621, setting a grid density discrete threshold for the clustering grid density discrete coefficient GDDC ;

[0303] Step S3622: Set the asset density discrete threshold for the cluster asset density discrete coefficient ADDC ;

[0304] Step S3623, setting the intra-cluster grid size variation threshold for the intra-cluster grid size variation coefficient GGVC ;

[0305] Step S3624, setting the inter-cluster grid size variation coefficient GGVB for the inter-cluster grid size variation coefficient GGVB ;

[0306] Step S3625: When two or more retrieval quality evaluation parameters exceed their respective thresholds, a second condition for triggering dynamic grid adjustment is determined.

[0307] Example 3

[0308] This embodiment discloses an electronic device 500, such as Fig.10 As shown, the electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, which, when executed by one or more processors, can execute the above-mentioned spatial index-based individual loan non-performing asset grid management geographic information retrieval method.

[0309] The method or system according to the embodiment of the present application can also be used Fig.10 The electronic device architecture shown in FIG. Fig.10 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store the spatial index-based individual loan non-performing asset grid management geographic information retrieval method provided in this application. The geographic information retrieval method for grid management of personal loan non-performing assets based on spatial index may, for example, include: performing initial grid division on the target area to form a first grid set, calculating the non-performing asset density of each grid in the first grid set to obtain a first density distribution; judging whether the first condition for dynamic grid adjustment is triggered based on the first density distribution; if triggered, performing adaptive re-division on the first grid set to generate a second grid set; if not triggered, maintaining the first grid set; constructing a first spatial index tree for the first grid set; obtaining an incremental grid set based on the first grid set and the second grid set; constructing an incremental spatial index tree based on the incremental grid set; merging the first spatial index tree with the incremental spatial index tree to form a second spatial index tree; receiving a geographic information retrieval request, performing a retrieval based on the second spatial index tree, and generating a retrieval result; judging whether the second condition for dynamic grid adjustment is triggered based on the retrieval result; if triggered, dynamically adjusting the first grid granularity parameter and the first grid shape parameter based on the retrieval quality evaluation parameter, and re-performing initial grid division on the target area; if not triggered, completing this retrieval and outputting the retrieval result.

[0310] Furthermore, the electronic device 500 may also include a user interface 508. Fig.10 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Fig.10 One or more components of an electronic device are shown.

[0311] Example 4

[0312] This embodiment discloses a computer-readable storage medium 600, such as Fig.11 As shown, computer-readable instructions are stored on the computer-readable storage medium 600. When the computer-readable instructions are executed by the processor, the geographic information retrieval method for grid management of individual loan non-performing assets based on spatial index according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium 600 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache). Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0313] In addition, according to the implementation mode of the present application, the process described in the above reference flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: initially gridding the target area to form a first grid set, calculating the non-performing asset density of each grid in the first grid set, and obtaining a first density distribution; judging whether to trigger the first condition of dynamic grid adjustment according to the first density distribution; if triggered, performing adaptive re-division on the first grid set to generate a second grid set; if not triggered, maintaining the first grid set; for the first grid set ... A grid set is used to construct a first spatial index tree; an incremental grid set is obtained according to the first grid set and the second grid set; an incremental spatial index tree is constructed according to the incremental grid set; the first spatial index tree is merged with the incremental spatial index tree to form a second spatial index tree; a geographic information retrieval request is received, and a retrieval is performed according to the second spatial index tree to generate a retrieval result; according to the retrieval result, it is determined whether the second condition of dynamic grid adjustment is triggered; if triggered, the first grid granularity parameter and the first grid shape parameter are dynamically adjusted according to the retrieval quality evaluation parameter, and the target area is re-initialized for grid division; if not triggered, the current retrieval is completed and the retrieval result is output. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0314] The methods, systems, and devices of the present application may be implemented in many ways. For example, the methods, systems, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0315] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0316] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A spatial index-based geographic information retrieval method for grid management of non-performing assets of individual loans, characterized by: The method comprises: Performing initial grid division on the target area to form a first grid set, calculating the non-performing asset density of each grid in the first grid set to obtain a first density distribution; judging whether a first condition for dynamic grid adjustment is triggered according to the first density distribution; if triggered, performing adaptive re-division on the first grid set to generate a second grid set; if not triggered, maintaining the first grid set; For the first grid set, construct a first spatial index tree; according to the first grid set and the second grid set, obtain an incremental grid set; according to the incremental grid set, construct an incremental spatial index tree; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree; Receive a geographic information retrieval request, perform retrieval according to the second spatial index tree, and generate retrieval results; determine whether the second condition of dynamic grid adjustment is triggered according to the retrieval results; if triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-divide the target area into initial grids; if not triggered, complete this retrieval and output the retrieval results; The initial grid division of the target area to form a first grid set includes: Obtain the geographic location information of individual loan non-performing assets, and unify the geographic location information into longitude and latitude coordinates in the WGS84 coordinate system; According to the preset first grid size parameter and the first grid shape parameter, the target area is initially grid-divided to form n1 grids; n1 is a positive integer; Extracting core attribute information of each grid; the core attribute information includes grid area, grid quantity, grid number, center coordinates and boundary range; Count the number of individual loan non-performing assets that fall into each grid, obtain the ID list of these individual loan non-performing assets, and establish a grid-asset mapping index; A first grid set is formed by core attribute information and grid-asset mapping index.

2. The method for geographic information retrieval based on spatial index for grid management of individual loan non-performing assets according to claim 1 is characterized in that: The step of calculating the non-performing asset density of each grid in the first grid set to obtain the first density distribution includes: traversing each grid in the first grid set, calculating the non-performing asset density of each grid by the ratio of the number of non-performing assets in the grid to the grid area, generating the first density distribution, and obtaining the distribution vector of the first density distribution. ; is the non-performing asset density of the i-th grid, i is the grid number in the first grid set, ; The first condition for determining whether to trigger dynamic grid adjustment according to the first density distribution includes: determining whether to trigger dynamic grid adjustment according to the first density distribution ... , calculating the discrete degree of the first density distribution; comparing the discrete degree of the first density distribution with a preset discrete threshold, and triggering the first condition for dynamic grid adjustment if the discrete degree of the first density distribution exceeds the discrete threshold.

3. The method for geographic information retrieval based on spatial index for grid management of individual loan non-performing assets according to claim 1 is characterized in that: The performing adaptive re-division on the first grid set to generate the second grid set comprises: Calculate the upper quartile and the lower quartile of the first density distribution, use the upper quartile as the first density threshold of the first density distribution, and use the lower quartile as the second density threshold of the first density distribution; Traversing the first grid set, dividing the grids whose non-performing asset density is higher than the first density threshold into four equal parts, and dividing them into four sub-grids to form a first sub-grid set; Traversing the first grid set, merging the grids whose non-performing asset density is lower than the second density threshold, and merging two grids into one to form a mother grid, thereby obtaining a first mother grid set; The first sub-grid set, the first mother grid set and the grids in the first grid set that have not been adaptively re-divided are integrated to obtain an adjusted grid set, and all grids in the adjusted grid set are numbered to form a second grid set.

4. The method for retrieval of geographic information of grid management of individual loan non-performing assets based on spatial index according to claim 3 is characterized in that: The method for numbering all the grids in the adjusted grid set comprises: Mark the grids in the first grid set that performs the quartering as the quartered original grids, hide the quartered original grids, and add two subscripts to the numbers of the four newly added subgrids on the basis of the numbers of the quartered original grids to obtain the grid numbers of all the subgrids in the first subgrid set; For each mother grid formed by combining two grids into one, the grid number of all mother grids in the first mother grid set is obtained by taking the grid number with the smaller lexicographic order of the original numbers of the two merged grids; For the grids in the first grid set that do not perform adaptive re-division, their numbers remain unchanged; Summarize the grid numbers of all subgrids in the first subgrid set, the grid numbers of all mother grids in the first mother grid set, and the grid numbers of grids that have not been adaptively repartitioned in the first grid set to obtain the grid numbers of all grids in the adjusted grid set; All grids in the adjusted grid set are renumbered according to the lexicographic order of the grid numbers; the lexicographic order of the grid numbers includes an increasing lexicographic order and a decreasing lexicographic order.

5. The method for geographic information retrieval based on spatial index for grid management of individual loan non-performing assets according to claim 1, characterized in that: The step of obtaining an incremental grid set according to the first grid set and the second grid set comprises: Traversing the second grid set, extracting grids whose numbers do not exist in the first grid set, to form a newly added grid set; Traversing the first grid set, extracting grids whose numbers do not exist in the second grid set, to form a deleted grid set; Traversing the second grid set, extracting the grids whose numbers are the same as those of a grid in the first grid set but whose core attribute information has changed, to form an updated grid set; The newly added grid set, deleted grid set, and updated grid set are unioned to obtain the incremental grid set.

6. The method for retrieval of geographic information of grid management of individual loan non-performing assets based on spatial index according to claim 1 is characterized in that: The geographic information search request includes the geographic location of the target individual loan non-performing assets, the target search range parameter and the target grid granularity parameter; The performing of the search and generating the search results comprises: Extracting a target grid granularity parameter as a second grid granularity parameter, comparing the second grid granularity parameter with the first grid granularity parameter, and determining whether the second grid granularity parameter is consistent with the first grid granularity parameter; If the second grid granularity parameter is consistent with the first grid granularity parameter, the search is directly performed using the second spatial index; If the second grid granularity parameter is inconsistent with the first grid granularity parameter, the second grid set is re-divided according to the second grid granularity parameter to form a third grid set, and a third spatial index tree corresponding to the third grid set is established, and the third spatial index tree is used to perform retrieval.

7. The method for retrieval of geographic information of grid management of individual loan non-performing assets based on spatial index according to claim 6 is characterized in that: The performing of the search and generating the search results further comprises: According to the target search range parameters, determine the spatial boundary of the rectangular search area centered on the geographical location of the target non-performing assets; Calculate the grid numbers of the starting grid and the ending grid covered by the rectangular search area, and use them as the search starting grid number and the search ending grid number respectively; Using the second spatial index tree or the third spatial index tree, traverse the grid set covered by the search start grid number to the search end grid number to form a candidate grid set; The geographical information of non-performing assets contained in the candidate grid set is extracted to generate a search result.

8. The method for geographic information retrieval based on spatial index for grid management of individual loan non-performing assets according to claim 1, characterized in that: The second condition for determining whether to trigger dynamic grid adjustment includes: Perform cluster analysis on the search results to generate search quality evaluation parameters that reflect the rationality of grid division within the time and space range; the search quality evaluation parameters include cluster grid density dispersion coefficient GDDC, cluster asset density dispersion coefficient ADDC, intra-cluster grid granularity variation coefficient GGVC, and inter-cluster grid granularity variation coefficient GGVB; Set the grid density discreteness threshold for the clustering grid density discreteness coefficient GDDC ; Set the asset density discrete threshold for the cluster asset density discrete coefficient ADDC ; Set the intra-cluster grid size variation threshold for the intra-cluster grid size variation coefficient GGVC ; Set the inter-cluster grid size variation coefficient GGVB to set the inter-cluster grid size variation ; When two or more retrieval quality evaluation parameters exceed their respective thresholds, a second condition for triggering dynamic grid adjustment is determined.

9. A spatial index-based individual loan non-performing assets grid management geographic information retrieval system, which is used to implement the spatial index-based individual loan non-performing assets grid management geographic information retrieval method according to any one of claims 1 to 8, characterized in that: The system comprises: Grid division module: used to perform initial grid division on the target area to form a first grid set, calculate the non-performing asset density of each grid in the first grid set, and obtain a first density distribution; according to the first density distribution, determine whether the first condition for dynamic grid adjustment is triggered; if triggered, perform adaptive re-division on the first grid set to generate a second grid set; if not triggered, keep the first grid set; Spatial index module: used to construct a first spatial index tree for a first grid set; obtain an incremental grid set according to the first grid set and the second grid set; construct an incremental spatial index tree according to the incremental grid set; merge the first spatial index tree with the incremental spatial index tree to form a second spatial index tree; Retrieval module: used to receive geographic information retrieval requests, perform retrieval according to the second spatial index tree, and generate retrieval results; based on the retrieval results, determine whether the second condition for dynamic grid adjustment is triggered; if triggered, dynamically adjust the first grid granularity parameter and the first grid shape parameter according to the retrieval quality evaluation parameter, and re-initialize the grid division of the target area; if not triggered, complete this retrieval and output the retrieval results.

10. An electronic device comprising a memory, a central processing unit, and a computer program stored in the memory and executable on the central processing unit, characterized in that: When the central processing unit executes the computer program, the geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing as described in any one of claims 1-8 is implemented.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed, implements the geographic information retrieval method for grid management of individual loan non-performing assets based on spatial indexing as described in any one of claims 1 to 8.

Images