Mesh laying method and device, electronic equipment and computer storage medium

Abstract

This invention provides a method, apparatus, electronic device, and computer storage medium for network deployment. First, after gridding the target city, a radial feature vector is obtained for each target grid. Then, the radial feature vectors for each target grid are input into a pre-trained performance prediction model to obtain predicted performance data for deploying a network point in each target grid. Finally, based on a preset performance threshold, existing network point data for the target city, and the predicted performance data for each target grid, at least one core grid suitable for deploying new network points is selected from all target grids. Since the radial feature vector reflects population data, macroeconomic data, and industry data radiating outwards from the center point of the target grid within a preset radius, this application improves the efficiency and accuracy of new network point site selection without requiring manual site selection.

Description

Technical Field

[0001] This invention relates to the field of big data technology, and more specifically, to a method, apparatus, electronic device, and computer storage medium for network deployment. Background Technology

[0002] In industries such as finance, retail, and new energy infrastructure, establishing physical service outlets is crucial for enhancing customer experience and ensuring service quality. Taking the financial industry as an example, banks typically need to strategically distribute their branches across various cities to ensure balanced and accessible service coverage. A well-planned service network not only improves business service efficiency but also effectively controls operating costs and achieves optimal resource allocation.

[0003] Currently, the selection of offline outlets mostly relies on manual methods. Specifically, technical personnel need to conduct surveys, collect data, and conduct comprehensive assessments of urban areas, making subjective judgments based on this. However, this method has significant shortcomings: on the one hand, the manual site selection process is cumbersome, time-consuming, and inefficient, requiring a large investment of human resources; on the other hand, the site selection results are easily influenced by personal experience and subjective factors, lacking objective and systematic evaluation criteria, making it difficult to guarantee the reliability of the site selection. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, electronic device, and computer storage medium for network deployment, in order to solve the problems existing in the prior art.

[0005] The embodiments of the present invention can be implemented as follows: In a first aspect, the present invention provides a method for laying out distribution networks, comprising: After the target city is processed into a grid, the radiation feature vector corresponding to each target grid of the target city is obtained; the radiation feature vector reflects the population data, macroeconomic data and industry data within a preset radius radiating outward from the center point of the target grid. The radiation feature vectors corresponding to each target grid are input into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid. Based on a preset performance threshold, existing network data of the target city, and predicted performance data for each target grid, at least one core grid is selected from all target grids where new network points can be deployed.

[0006] In an optional implementation, the performance prediction model is trained in the following manner: After gridding multiple reference cities, population data and industry data for each reference grid within each reference city are obtained; Obtain the location and actual performance data of each existing network point in each reference city; Training samples are constructed for each existing network point based on its actual performance data and population data and industry data of each reference grid within a preset radius. The training samples include radiation feature vector samples and actual performance labels. The initial prediction model is trained using all training samples to obtain the performance prediction model.

[0007] In an optional implementation, the step of constructing training samples corresponding to each existing network point based on the actual performance data of each existing network point and the population data and industry data of each reference grid within a preset radius includes: The service area of ​​each existing branch is defined by a pre-defined radius around it. Based on the service range of each existing outlet, all existing outlets are divided into independent outlets with non-overlapping service ranges and non-independent outlets with overlapping service ranges. All reference grids are then divided into single-coverage grids and multi-coverage grids, and the outlet group corresponding to each multi-coverage grid is determined. The outlet group includes at least two non-independent outlets. For each independent service point, population data and industry data of each reference grid within the service range of the independent service point are integrated to obtain service characteristic data of the independent service point within its service range; For each multi-coverage grid, the population data of the multi-coverage grid is assigned to each non-independent point in the corresponding point group; For each non-independent branch, the population data and industry data of each single-coverage grid and the allocated population data and industry data of each multi-coverage grid within the service range of the non-independent branch are integrated to obtain the service characteristic data of the non-independent branch within its service range. The service characteristic data of each existing outlet within its service area and the macroeconomic data of the reference city where the existing outlet is located are combined into multi-dimensional characteristic data and then standardized to obtain the radiation characteristic vector sample of each existing outlet. By associating the radiation feature vector samples of each existing network point with the actual performance data, training samples corresponding to each existing network point are obtained.

[0008] In an optional implementation, the step of allocating the population data of the multi-coverage grid to each non-independent point in the corresponding point group includes: Determine the attractiveness attributes of each non-independent site within the site group; Based on the attractiveness attribute of each non-independent point in the point group and the distance between the multi-coverage grid and each non-independent point in the point group, the allocation weight of each non-independent point in the point group relative to the multi-coverage grid is calculated. According to the allocation weight corresponding to each non-independent point within the point group, the population data of the multi-coverage grid is allocated to each non-independent point.

[0009] In an optional implementation, the formula for calculating the allocation weight is:

[0010] In the formula, Indicates the first in the network group A non-independent branch, Indicates multiple coverage grids, The number of non-independent outlets within the outlet group; Indicates the first A non-independent point relative to a multi-coverage grid The allocation weights; , They represent the first The first non-independent branch, The attractiveness attributes of a non-independent branch; They represent the first The first non-independent branch, From individual non-independent points to multi-coverage grids The distance; Indicates the distance attenuation sensitivity coefficient; This represents the maximum reachable distance constraint threshold; This indicates an indicator function; its value is 1 if the condition within the parentheses is true, and 0 otherwise.

[0011] In an optional implementation, the existing network data includes the location and range overlap of each existing network point in the target city; The step of selecting at least one core grid from all target grids where new grids can be deployed, based on a preset performance threshold, the location of each existing grid point in the target city, and the predicted performance data corresponding to each target grid, includes: Based on the preset performance threshold, the existing network data of the target city, and the predicted performance data corresponding to each target grid, an incremental strategy is used to select at least one core grid from all target grids where new network points can be laid. And / or, Based on the preset performance threshold, the location and range repetition of each existing network point in the target city, and the predicted performance data corresponding to each target grid, at least one core grid that can be used to lay new network points is selected from all target grids using the stock adjustment strategy. And / or, Based on the preset performance threshold and the predicted performance data corresponding to each target grid, a full redigradation strategy is adopted to select the core grids that can be redigraded with new grid points from all target grids.

[0012] In an optional implementation, the step of selecting at least one core grid from all target grids where new grids can be deployed, based on a preset performance threshold, the location of each existing grid point within the target city, and the predicted performance data corresponding to each target grid, using an incremental strategy, includes: Remove all target grids within a preset radius around each existing grid point and all target grids whose predicted performance data is lower than the preset performance threshold from all target grids to obtain a grid list; The remaining target grids in the grid list are traversed sequentially. Each target grid is used as the core grid, and all target grids within a preset radius around the core grid are removed from the grid list until the grid list is empty.

[0013] In an optional implementation, the step of selecting at least one core grid from all target grids for new grid deployment based on the preset performance threshold, the location and range repetition of each existing grid point in the target city, and the predicted performance data corresponding to each target grid, using a stock adjustment strategy, includes: Each existing network point with a range repetition degree not lower than the repetition degree threshold is designated as a network point to be replaced, and each existing network point with a range repetition degree lower than the repetition degree threshold is designated as a network point to be retained. Remove all target grids within a preset radius around each retained grid point from all target grids, as well as all target grids whose predicted performance data is lower than the preset performance threshold, to obtain a grid list; The remaining target grids in the grid list are traversed sequentially. Each target grid is used as the core grid, and all target grids within a preset radius around the core grid are removed from the grid list until the grid list is empty or the sum of the number of the core grids and the number of retained grid points is not less than a set number.

[0014] In a second aspect, the present invention provides a network laying device, comprising: The feature extraction module is used to obtain the radiation feature vector corresponding to each target grid of the target city after the target city is gridded; the radiation feature vector reflects the population data, macroeconomic data and industry data within a preset radius radiating outward from the center point of the target grid. The prediction module is used to input the radiation feature vectors corresponding to each target grid into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid. The network point calculation module is used to select at least one core grid from all target grids where new network points can be laid, based on a preset performance threshold, existing network point data of the target city, and predicted performance data corresponding to each target grid.

[0015] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory stores a software program, and when the electronic device is running, the processor executes the software program to implement the method described in the first aspect above.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0017] Compared with existing technologies, this invention provides a network deployment method, apparatus, electronic device, and computer storage medium. First, after gridding the target city, a radiation feature vector is obtained for each target grid. This vector reflects population data, macroeconomic data, and industry data radiating outwards from the center point of the target grid within a preset radius. Next, the radiation feature vectors for each target grid are input into a pre-trained performance prediction model to obtain predicted performance data for deploying a new network point in each target grid. Finally, based on a preset performance threshold, existing network point data in the target city, and the predicted performance data for each target grid, at least one core grid suitable for deploying a new network point is selected from all target grids. This eliminates the need for manual site selection. Instead, it collects the radiation feature vectors of each target grid in the target city and inputs them into the performance prediction model to determine the predicted performance data for each target grid. Finally, by combining the preset performance threshold and existing network point data, at least one core grid suitable for deploying a new network point is selected from all target grids, improving the efficiency and accuracy of new network point deployment site selection. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is one of the flowcharts illustrating a network deployment method provided in an embodiment of the present invention.

[0020] Figure 2 This is a second schematic flowchart of a network deployment method provided in an embodiment of the present invention.

[0021] Figure 3 This is the third flowchart illustrating a network deployment method provided in an embodiment of the present invention.

[0022] Figure 4 This is a schematic diagram of the distribution of network points in a sample city, provided as an embodiment of the present invention.

[0023] Figure 5 This is a schematic diagram of a network laying device provided in an embodiment of the present invention.

[0024] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0026] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0027] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0028] Furthermore, the terms "first" and "second" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0029] It should be noted that, where there is no conflict, the features in the embodiments of the present invention can be combined with each other.

[0030] Please see Figure 1 , Figure 1 This is a flowchart illustrating a network deployment method according to an embodiment of the present invention. The method can be executed by computing devices such as smartphones, personal laptops, personal computers, and servers. The method includes the following steps S201 to S203.

[0031] S201. After the target city is processed into a grid, the radiation feature vector corresponding to each target grid of the target city is obtained.

[0032] In this embodiment, the target city is divided into several target grids. The radiation feature vector reflects population data, macroeconomic data, and industry data within a preset radius radiating outward from the center point of the target grid. These data can all affect the operational efficiency of the service points. The preset radius radiating outward from the center point of the target grid represents the service area for which a service point is to be established within the target grid.

[0033] Optionally, the target city can be any city. The gridding process uses Geohash technology to divide the target city into multiple approximately rectangular GeoHash grids. The specific processing procedure is an existing technology and will not be described in detail here.

[0034] The preset radius can be flexibly set by taking into account factors such as the size of the target city and its level of economic development. For example, the preset radius can be 1km or 2km.

[0035] S202. Input the radiation feature vectors corresponding to each target grid into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid.

[0036] In this embodiment, the radiation feature vector corresponding to a target grid is input into the performance prediction model to obtain a predicted performance data, which can reflect the predicted operating performance value that can be achieved after a network point is to be laid in the target grid. The predicted value can reflect the probability strength of the target grid location supporting the continuous operation of a single network point and achieving a reasonable performance level under the current economic and demographic structure.

[0037] S203. Based on the preset performance threshold, the existing network data of the target city, and the predicted performance data corresponding to each target grid, select at least one core grid from all target grids where new network points can be laid.

[0038] In this embodiment, new nodes can be deployed in the core grid area. The final core grid is a set of discrete locations selected from the remaining target grids after removing areas already effectively covered by existing nodes and areas below a preset performance threshold. These locations are ranked high in predicted performance data and their service areas do not overlap.

[0039] The network deployment method provided in this invention first obtains the radiation feature vector corresponding to each target grid in the target city after gridding. Then, the radiation feature vectors corresponding to each target grid are input into a pre-trained performance prediction model to obtain the predicted performance data achievable by deploying a network point in each target grid. Finally, based on a preset performance threshold, existing network point data in the target city, and the predicted performance data corresponding to each target grid, at least one core grid suitable for deploying new network points is selected from all target grids. Since the radiation feature vector reflects population data, macroeconomic data, and industry data radiating outward from the center point of the target grid within a preset radius, this application can improve the efficiency and accuracy of new network point deployment site selection without manual site selection.

[0040] Optionally, the network deployment method provided by this invention can be applied to, but is not limited to, the following industries: (1) Financial industry: Location selection of bank branches; (2) Retail industry: Site selection for new stores such as chain supermarkets, convenience stores, and pharmacies; (3) New energy infrastructure industry: site selection for charging stations, battery swapping stations, etc.

[0041] The following section uses bank branch location selection as an example to provide a detailed explanation of the implementation process of the branch network deployment method provided in this application.

[0042] In step S201 above, macroeconomic data for the target city can be collected first, such as per capita savings coefficient and per capita income. Next, for each target grid, the process of obtaining the corresponding radiation feature vector can be as follows: (1) Assuming that a grid point (called the proposed grid point) is to be set up at the center point of the target grid, its service range is the range of the grid center point radiating outward with a preset radius. It is necessary to collect population data and industry data within the service range of the proposed grid point first. (2) Combine the population data and industry data within the service area of ​​the proposed network point with the macroeconomic data of the target city, and perform standardization processing to obtain the radiation feature vector of the target grid. Standardization processing can eliminate the influence of dimensional differences on the model prediction results.

[0043] Among them, population data can include dimensions such as total population, age distribution ratio (such as the population ratio or number of people in the youth, middle-aged and elderly groups), income distribution ratio (such as the population ratio or number of people in the low-income, middle-income and high-income groups), and education distribution ratio (the population ratio or number of people in different education levels); industry type data reflects the distribution of industry types, which can include the number of competitor outlets and the degree of overlap in service scope with other existing outlets.

[0044] Taking Bank A's need to expand its branch network as an example, the number of competing branches represents the number of branches of other banks besides Bank A within the service area. The degree of overlap in service areas can have two scenarios: (1) The overlap rate between the service area of ​​the proposed branch and the service area of ​​the existing branch of nearby bank A (i.e., the ratio of the size of the overlapping area to the size of the service area of ​​the proposed branch). (2) The overlap rate between the service area of ​​the proposed branch and the service area of ​​existing branches of other banks in the vicinity (i.e., the ratio of the size of the overlapping area to the size of the service area of ​​the proposed branch).

[0045] Next, we will introduce the training process of the performance prediction model used in step S202.

[0046] In one optional implementation, training samples are constructed based on the existing network points in multiple reference cities to train the performance prediction model. See [link to relevant documentation]. Figure 2 The performance prediction model can be trained through the following steps S101~S104: S101. After performing gridding on multiple reference cities, obtain population data and industry data for each reference grid within each reference city.

[0047] In this embodiment, each reference city also needs to be gridded to divide it into several reference grids. However, unlike step S201 above, here we collect population data and industry data within the region of each reference grid, and the specific data content of the population data and industry data is described above.

[0048] For example, reference cities may include Beijing, Shanghai, Qingdao, Hangzhou, Guangzhou, Chengdu, Shenzhen, etc. The target city may be one of the reference cities or another city, which is not limited here.

[0049] S102. Obtain the location and actual performance data of each existing network point in each reference city.

[0050] The actual performance data can be an operational performance value calculated by taking into account multiple factors of the existing network.

[0051] S103. Based on the actual performance data of each existing network point and the population data and industry data of each reference grid within a preset radius, construct training samples corresponding to each existing network point.

[0052] The training samples include radiation feature vector samples and actual performance labels.

[0053] S104. Train the initial prediction model using all training samples to obtain the performance prediction model.

[0054] In this embodiment, all training samples constitute a dataset, which can be divided into a training set, a validation set, and a test set according to a certain ratio to achieve training, validation, and testing. In this way, the trained performance prediction model can quantitatively estimate the potential operational efficiency of a proposed network point in any grid within any city.

[0055] The initial prediction model can be, but is not limited to, regression models such as XGBoost (eXtreme Gradient Boosting) and MLP (Multilayer Perceptron).

[0056] In step S103 above, for any existing network point in any reference city, it is necessary to integrate the information of all reference grids within the service range of the existing network point to construct training samples. However, some of these reference grids are only within the service range of a single existing network point, while others are within the service range of multiple existing network points.

[0057] For example, within a three-kilometer radius of a large residential community, there are three branches (A, B, and C). The community's population structure, consumption levels, and commercial amenities should all influence the operational performance of these three branches. However, if the training samples from all three branches are included repeatedly, the model will misjudge the contribution of these features to any single branch. Conversely, simply assigning or averaging the data fails to reflect the actual differences in the attractiveness of each branch. For instance, branch A, with its larger area, more staff, and stronger brand recognition, should rightfully receive a higher weight in its regional characteristic attribution than branches B and C. Therefore, it is necessary to assign population data from the areas covered by multiple branches to ensure the accuracy of the constructed training samples.

[0058] That is, in Figure 2 Based on this, please refer to Figure 3 The implementation process of the above step S103 may include the following sub-steps S1031 to S1037.

[0059] S1031. The pre-defined radius around each existing branch is used as its service area.

[0060] S1032. Based on the service range of each existing outlet, all existing outlets are divided into independent outlets with non-overlapping service ranges and non-independent outlets with overlapping service ranges. All reference grids are then divided into single-coverage grids and multi-coverage grids, and the outlet group corresponding to each multi-coverage grid is determined.

[0061] In this embodiment, a group of dots corresponding to a multi-coverage grid includes at least two non-independent dots.

[0062] For example, Figure 4 This shows the locations and service areas of four existing service points (i.e., service points A, B, C, and D) in a sample city. Figure 4 Branches A, B, and C are all non-independent branches, while only branch D is an independent branch.

[0063] Because the service areas of points A, B, and C overlap, all 12 grids filled with red diagonal stripes are multi-coverage grids, and the corresponding point groups for each of these 12 grids include points A and B. Similarly, the 4 grids filled with blue diagonal stripes are also multi-coverage grids, but the corresponding point groups for each of these 4 grids include points A, B, and C. In addition to these 16 multi-coverage grids, Figure 4 The remaining grids shown are all single-coverage grids.

[0064] It should be noted that this example is merely illustrative and is not intended to be limiting.

[0065] S1033. For each independent branch, integrate the population data and industry data of each reference grid within the service range of the independent branch to obtain the service characteristic data of the independent branch within its service range.

[0066] In this embodiment, for an independent service point, the population data and industry data of each reference grid within the service range of the independent service point can be summed according to the dimension to obtain the integrated service feature data corresponding to the independent service point.

[0067] For example, assuming population data involves 11 dimensions and industry data involves 2 dimensions, then for Figure 4In the example of point D, whose service area includes 52 grids, the service feature data of point D within its service area is obtained by summing the values ​​of the 52 grids within each of the 12 dimensions for each dimension. This integration is the culmination of the summation of the values ​​for each of the 52 grids within point D's service area. It should be noted that this example is merely illustrative and is not intended to be limiting.

[0068] S1034. For each multi-coverage grid, the population data of the multi-coverage grid is assigned to each non-independent grid in the corresponding grid group.

[0069] In this embodiment, for a multi-coverage grid, after the population data of the multi-coverage grid is allocated to each non-independent point in the corresponding point group, each non-independent point in the point group will correspond to an allocated population data in the multi-coverage grid.

[0070] Optionally, population data for multiple coverage grids can be obtained by calculating the weights of each non-independent grid within the grid group. That is, the implementation process of step S1034 may include steps a1 to a3.

[0071] Step a1: Determine the attractiveness attributes of each non-independent outlet within the outlet group.

[0072] For example, in the application scenario of selecting bank branch locations, attractiveness attributes can be comprehensively determined by dimensions such as branch area, staffing ratio, and number of counters. Attractiveness attributes characterize the branch's appeal to attract users to conduct business there. It should be noted that this example is merely illustrative and not intended to limit the scope; in other application scenarios, the dimensions considered for attractiveness attributes can be set according to the actual situation.

[0073] Step a2: Based on the attraction attributes of each non-independent point in the point group and the distance between the multi-coverage grid and each non-independent point in the point group, calculate the weight of each non-independent point in the point group relative to the multi-coverage grid.

[0074] In addition to the attractiveness attributes of each non-independent point within the point group and the distance of each non-independent point to the multi-coverage grid, the weight of each non-independent point in the multi-coverage grid can be determined by combining the maximum reachable distance constraint threshold and the distance decay sensitivity coefficient. This will enable the accurate allocation of various dimensions of population data in the multi-point coverage area, thereby solving the problem of ambiguous feature attribution in the multi-point shared coverage grid.

[0075] Assuming the multi-coverage grid is the first in a sample city Given a reference grid whose corresponding dot group contains M non-independent dots, then for the i-th reference grid... For a non-independent network point, the formula for calculating its assigned weight can be:

[0076] In the formula, Indicates multi-coverage grid The first in the corresponding network group A non-independent branch, Indicates multiple coverage grids, The number of non-independent outlets within the outlet group; Indicates the first A non-independent point relative to a multi-coverage grid The allocation weights; , They represent the first The first non-independent branch, The attractiveness attributes of a non-independent branch; They represent the first The first non-independent branch, From individual non-independent points to multi-coverage grids The distance; Indicates the distance attenuation sensitivity coefficient. It can be set to 0.6 or 0.8; This represents the maximum reachable distance constraint threshold. It can be set to 1km, 1.5km, or 2km, etc.; This indicates an indicator function; its value is 1 if the condition within the parentheses is true, and 0 otherwise.

[0077] As can be seen from the formula for calculating the weight allocation, once the first... From individual non-independent points to multi-coverage grids The distance exceeds ,So =0.

[0078] Step a3: According to the allocation weight of each non-independent point in the point group, allocate the population data of the multi-coverage grid to each non-independent point.

[0079] For multi-coverage grids The first in the corresponding network group Each non-independent point of view multiplies each dimension of the population data from the multi-coverage grid by 1. That is, to obtain the first Multiple grid coverage areas within the service area of ​​each non-independent branch The allocated population data.

[0080] S1035. For each non-independent branch, integrate the population data and industry data of each single-coverage grid and the allocated population data and industry data of each multi-coverage grid within the service range of the non-independent branch to obtain the service characteristic data of the non-independent branch within its service range.

[0081] For example, assuming population data involves 11 dimensions and industry data involves 2 dimensions, then for Figure 4 For example, if network point A has 16 multi-coverage grids and 35 single-coverage grids within its service area, then for each of the 12 dimensions, the values ​​of the 16 multi-coverage grids and 35 single-coverage grids within network point A's service area are summed to obtain the service characteristic data of network point A within its service area. It should be noted that this example is merely illustrative and is not intended to limit the scope of the analysis.

[0082] S1036. After combining the service characteristic data of each existing outlet within its service area and the macroeconomic data of the reference city where the existing outlet is located into multi-dimensional characteristic data, the data is standardized to obtain the radiation characteristic vector sample of each existing outlet.

[0083] S1037. Associate the radiation feature vector samples of each existing network point with the actual performance data to obtain the training samples corresponding to each existing network point.

[0084] In this embodiment, the construction of radiation feature vector samples also takes into account the macroeconomic data of the reference city (such as per capita deposit coefficient, per capita income, etc.). The radiation feature vector sample of an existing branch is associated with the actual performance data (as the actual performance label) to obtain the training sample corresponding to the existing branch.

[0085] For step S203 above, the existing network data includes the location and range overlap of each existing network point in the target city. Therefore, step S203 can be implemented by at least one of the following steps S2031 to S2033.

[0086] S2031. Based on the preset performance threshold, the existing network data of the target city and the predicted performance data corresponding to each target grid, an incremental strategy is used to select at least one core grid from all target grids where new network points can be laid. S2032. Based on the preset performance threshold, the location and range repetition of each existing network point in the target city, and the predicted performance data corresponding to each target grid, at least one core grid that can be used to lay new network points is selected from all target grids using the stock adjustment strategy. S2033. Based on the preset performance threshold and the predicted performance data corresponding to each target grid, a full redipping strategy is adopted to select each core grid that can be redipping new grid points from all target grids.

[0087] The preset performance threshold can be flexibly set based on the desired performance target and the development status of the target city. Taking Shanghai as the target city as an example, the preset performance thresholds for high, low, and medium expectations can be 4.88, 3.27, and 2.89, respectively; while for Changsha as the target city, the preset performance thresholds for high, low, and medium expectations can be 1.71, 0.75, and 2.05, respectively. It should be noted that this example is for illustrative purposes only and is not intended to be limiting.

[0088] In other words, this application can employ at least one of the following three location selection strategies to filter the core grid: (1) Incremental strategy: Based on retaining all existing network points in the target city, in areas not covered by network points, the predicted performance data of each target grid and the distance constraints of surrounding network points are used to screen core grids that meet the performance standards for the purpose of laying new network points. (2) Stock adjustment strategy: Only existing outlets that meet the performance standards are retained in all existing outlets in the target city. Then, in areas not covered by outlets, areas covered by existing outlets that do not meet the performance standards, and areas with a high proportion of overlapping coverage, core grids that meet the performance standards are selected for the purpose of laying new outlets. (3) Full redeployment strategy: Simulate laboratory scenario, completely disregard all existing network points in the target city, and select sites without blind spots in the target city area, select core grids that meet the efficiency standards for laying new network points, and realize the redeployment of network points throughout the city.

[0089] In an optional implementation, if a stock adjustment strategy is adopted, the implementation process of step S2031 above may include steps b1 to b2.

[0090] Step b1: Remove all target grids within a preset radius around each existing grid point and all target grids whose predicted performance data is lower than a preset performance threshold from all target grids to obtain a grid list; Step b2: Iterate through each remaining target grid in the grid list, using each target grid as the core grid and removing all target grids within a preset radius around the core grid from the grid list until the grid list is empty.

[0091] In an optional implementation, when a stock adjustment strategy is adopted, the implementation process of step S2032 above may include steps c1 to c3.

[0092] Step c1: Select each existing point with a range repetition degree not lower than the repetition degree threshold as the point to be replaced, and select each existing point with a range repetition degree lower than the repetition degree threshold as the point to be retained. Step c2: Remove all target grids within a preset radius around each retained grid point and all target grids whose prediction performance data is lower than a preset performance threshold from all target grids to obtain a grid list; Step c3: Iterate through each remaining target grid in the grid list, take the target grid as the core grid, and remove all target grids within a preset radius around the core grid from the grid list until the grid list is empty or the sum of the number of core grids and the number of retained grid points is not less than the set number.

[0093] The set quantity can be the number of existing outlets in the target city, or it can be a newly set value, such as 300 or 200.

[0094] In the optional implementation method, when a full re-tiling strategy is adopted, the implementation process of step S2033 above can be as follows: Remove all target grids whose predicted performance data is lower than a preset performance threshold from all target grids to obtain a grid list; sequentially traverse each remaining target grid in the grid list, take that target grid as the core grid, and remove all target grids within a preset radius around the core grid from the grid list, until the grid list is empty.

[0095] The three location selection strategies provided by this invention can respectively meet the location selection requirements under different business constraint scenarios.

[0096] It should be noted that the execution order of each step in the above method embodiments is not limited to that shown in the accompanying drawings, and the execution order of each step shall be subject to the actual application. In this application, the terms "existing network point" (belonging to the target city) and "current network point" (belonging to the sample city) have the same meaning, and are only used to facilitate the distinction between the target city and the sample city.

[0097] In order to perform the corresponding steps in the above method embodiments and various possible implementations, an implementation method of a network laying device is given below.

[0098] Please see Figure 5 , Figure 5 A schematic diagram of the dot-laying device provided in an embodiment of the present invention is shown. The dot-laying device 200 includes: a feature extraction module 210, a prediction module 220, and a dot calculation module 230.

[0099] The feature extraction module 210 is used to obtain the radiation feature vector corresponding to each target grid of the target city after the target city is gridded; the radiation feature vector reflects the population data, macroeconomic data and industry data within a preset radius radiating outward from the center point of the target grid. The prediction module 220 is used to input the radiation feature vectors corresponding to each target grid into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid. The network point calculation module 230 is used to select at least one core grid from all target grids that can be used to lay new network points, based on a preset performance threshold, existing network point data of the target city and predicted performance data corresponding to each target grid.

[0100] Optionally, the network deployment device 200 may also include a training module, which can be used to train the performance prediction model in the following ways: after gridding multiple reference cities, obtain population data and industry data for each reference grid in each reference city; obtain the location and actual performance data of each existing network point in each reference city; construct training samples corresponding to each existing network point based on the actual performance data of each existing network point and the population data and industry data of each reference grid within a preset radius; the training samples include radiation feature vector samples and actual performance labels; and train the initial prediction model using all training samples to obtain the performance prediction model.

[0101] Optionally, in the process of "constructing training samples corresponding to each existing network point based on the actual performance data of each existing network point and the population data and industry data of each reference grid within a preset radius," the training module can specifically be used for: The service area of ​​each existing branch is defined by a pre-defined radius around it. Based on the service range of each existing outlet, all existing outlets are divided into independent outlets with non-overlapping service ranges and non-independent outlets with overlapping service ranges. All reference grids are then divided into single-coverage grids and multi-coverage grids, and the outlet group corresponding to each multi-coverage grid is determined. The outlet group includes at least two non-independent outlets. For each independent branch, population data and industry data of each reference grid within the service area of ​​the independent branch are integrated to obtain service characteristic data of the independent branch within its service area; For each multi-coverage grid, the population data of the multi-coverage grid is assigned to each non-independent point in the corresponding point group; For each non-independent branch, the population data and industry data of each single-coverage grid and the allocated population data and industry data of each multi-coverage grid within the service range of the non-independent branch are integrated to obtain the service characteristic data of the non-independent branch within its service range. After combining the service characteristic data of each existing outlet within its service area with the macroeconomic data of the reference city where the existing outlet is located into multidimensional feature data, the data is standardized to obtain the radiation feature vector sample of each existing outlet. By associating the radiation feature vector samples of each existing network point with the actual performance data, training samples corresponding to each existing network point are obtained.

[0102] Optionally, in the process of "assigning population data from a multi-coverage grid to each non-independent point in the corresponding point group", the training module can specifically be used to: determine the attractiveness attribute of each non-independent point in the point group; calculate the allocation weight of each non-independent point in the point group relative to the multi-coverage grid based on the attractiveness attribute of each non-independent point in the point group and the distance between the multi-coverage grid and each non-independent point in the point group; and allocate the population data of the multi-coverage grid to each non-independent point according to the allocation weight corresponding to each non-independent point in the point group.

[0103] Optionally, the existing network data includes the location and extent overlap of each existing network point in the target city. The network calculation module 230 can specifically be used for: Based on preset performance thresholds, existing network data for the target city, and predicted performance data for each target grid, an incremental strategy is used to select at least one core grid from all target grids where new network points can be deployed; and / or, Based on preset performance thresholds, the location and extent redundancy of each existing grid point within the target city, and the predicted performance data corresponding to each target grid, a stock adjustment strategy is used to select at least one core grid from all target grids where new grid points can be deployed; and / or, Based on the preset performance threshold and the predicted performance data corresponding to each target grid, a full redigradation strategy is adopted to select the core grids that can be redigraded with new grid points from all target grids.

[0104] Optionally, the network point calculation module 230, in the process of "selecting at least one core grid from all target grids where new network points can be deployed based on a preset performance threshold, the location of each existing network point in the target city, and the predicted performance data corresponding to each target grid, using an incremental strategy," can specifically be used for: Remove all target grids within a preset radius around each existing grid point and all target grids whose predicted performance data is lower than a preset performance threshold from all target grids to obtain a grid list; iterate through each remaining target grid in the grid list, take the target grid as the core grid and remove all target grids within a preset radius around the core grid from the grid list, until the grid list is empty.

[0105] Optionally, in the process of "selecting at least one core grid from all target grids for new grid deployment based on a stock adjustment strategy, using a preset performance threshold, the location and range repetition of each existing grid point in the target city, and the predicted performance data corresponding to each target grid," the grid calculation module 230 can specifically be used to: designate each existing grid point with a range repetition not lower than the repetition threshold as a grid point to be replaced, and designate each existing grid point with a range repetition lower than the repetition threshold as a retained grid point; remove each target grid within a preset radius around each retained grid point and each target grid with predicted performance data lower than the preset performance threshold from all target grids to obtain a grid list; sequentially traverse each remaining target grid in the grid list, designate the target grid as a core grid, and remove each target grid within a preset radius around the core grid from the grid list until the grid list is empty or the sum of the number of core grids and retained grid points is not lower than a set number.

[0106] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the network laying device 200 described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0107] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. The electronic device 300 includes a processor 310, a memory 320, and a bus 330, with the processor 310 connected to the memory 320 via the bus 330.

[0108] The memory 320 can be used to store software programs or firmware, for example, the software program or firmware corresponding to the above-mentioned network laying device 200. The processor 310 executes various functional applications and data processing by running the software program stored in the memory 320 to realize the network laying method provided in the embodiments of the present invention.

[0109] The memory 320 may be, but is not limited to, RAM (Random Access Memory), ROM (Read Only Memory), FLASH (Flash Memory), PROM (Programmable Read-Only Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electric Erasable Programmable Read-Only Memory), etc.

[0110] The processor 310 can be an integrated circuit chip with signal processing capabilities, capable of executing software programs, such as the software program corresponding to the aforementioned network laying device 200. The processor 310 can be a general-purpose processor, including: CPU (Central Processing Unit), NP (Network Processor), SoC (System on Chip), etc.; it can also be: DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0111] Understandable. Figure 6 The structure shown is for illustrative purposes only; the electronic device 300 may also include components that are more advanced than those shown. Figure 6 The more or fewer components shown, or having the same Figure 6 The different configurations shown. Figure 6 The components shown can be implemented using hardware, software, or a combination thereof.

[0112] This invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the dot-matrix laying method disclosed in the above embodiments. The computer-readable storage medium can be, but is not limited to, various media capable of storing program code, such as a USB flash drive, portable hard drive, ROM, RAM, PROM, EPROM, EEPROM, FLASH, magnetic disk, or optical disk.

[0113] This invention also provides a computer program product that, when executed by a processor, implements the network laying method disclosed in the above embodiments.

[0114] In summary, this invention provides a network deployment method, apparatus, electronic device, and computer storage medium. First, after gridding the target city, a radial feature vector is obtained for each target grid. This radial feature vector reflects population data, macroeconomic data, and industry data radiating outwards from the center point of the target grid within a preset radius. Next, the radial feature vectors for each target grid are input into a pre-trained performance prediction model to obtain predicted performance data for deploying a new network point in each target grid. Finally, based on a preset performance threshold, existing network point data in the target city, and the predicted performance data for each target grid, at least one core grid suitable for deploying a new network point is selected from all target grids. This eliminates the need for manual site selection. Instead, it collects the radial feature vectors of each target grid in the target city and inputs them into the performance prediction model to determine the predicted performance data for each target grid. Finally, by combining the preset performance threshold and existing network point data of the target city, at least one core grid suitable for deploying a new network point is selected from all target grids, improving the efficiency and accuracy of new network point deployment site selection.

[0115] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for laying network points, characterized in that, include: After gridding the target city, the radiation feature vector corresponding to each target grid of the target city is obtained; The radiation feature vector reflects population data, macroeconomic data, and industry data radiating outward from the center point of the target grid within a preset radius. The radiation feature vectors corresponding to each target grid are input into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid. Based on a preset performance threshold, existing network data of the target city, and predicted performance data for each target grid, at least one core grid is selected from all target grids where new network points can be deployed.

2. The method according to claim 1, characterized in that, The performance prediction model is trained in the following way: After gridding multiple reference cities, population data and industry data for each reference grid within each reference city are obtained; Obtain the location and actual performance data of each existing network point in each reference city; Training samples are constructed for each existing network point based on its actual performance data and population data and industry data of each reference grid within a preset radius. The training samples include radiation feature vector samples and actual performance labels. The initial prediction model is trained using all training samples to obtain the performance prediction model.

3. The method according to claim 2, characterized in that, The step of constructing training samples corresponding to each existing network point based on the actual performance data of each existing network point and the population data and industry data of each reference grid within a preset radius includes: The service area of ​​each existing branch is defined by a pre-defined radius around it. Based on the service range of each existing outlet, all existing outlets are divided into independent outlets with non-overlapping service ranges and non-independent outlets with overlapping service ranges. All reference grids are then divided into single-coverage grids and multi-coverage grids, and the outlet group corresponding to each multi-coverage grid is determined. The outlet group includes at least two non-independent outlets. For each independent service point, population data and industry data of each reference grid within the service range of the independent service point are integrated to obtain service characteristic data of the independent service point within its service range; For each multi-coverage grid, the population data of the multi-coverage grid is assigned to each non-independent point in the corresponding point group; For each non-independent branch, the population data and industry data of each single-coverage grid and the allocated population data and industry data of each multi-coverage grid within the service range of the non-independent branch are integrated to obtain the service characteristic data of the non-independent branch within its service range. The service characteristic data of each existing outlet within its service area and the macroeconomic data of the reference city where the existing outlet is located are combined into multi-dimensional characteristic data and then standardized to obtain the radiation characteristic vector sample of each existing outlet. By associating the radiation feature vector samples of each existing network point with the actual performance data, training samples corresponding to each existing network point are obtained.

4. The method according to claim 3, characterized in that, The step of allocating the population data of the multi-coverage grid to each non-independent point in the corresponding point group includes: Determine the attractiveness attributes of each non-independent site within the site group; Based on the attractiveness attribute of each non-independent point in the point group and the distance between the multi-coverage grid and each non-independent point in the point group, the allocation weight of each non-independent point in the point group relative to the multi-coverage grid is calculated. According to the allocation weight corresponding to each non-independent point within the point group, the population data of the multi-coverage grid is allocated to each non-independent point.

5. The method according to claim 4, characterized in that, The formula for calculating the allocation weights is as follows: In the formula, Indicates the first in the network group A non-independent branch, Indicates multiple coverage grids, The number of non-independent outlets within the outlet group; Indicates the first A non-independent point relative to a multi-coverage grid The allocation weights; , They represent the first The first non-independent branch, The attractiveness attributes of a non-independent branch; They represent the first The first non-independent branch, From individual non-independent points to multi-coverage grids The distance; Indicates the distance attenuation sensitivity coefficient; This represents the maximum reachable distance constraint threshold; This indicates an indicator function; its value is 1 if the condition within the parentheses is true, and 0 otherwise.

6. The method according to claim 1, characterized in that, The existing network data includes the location and extent redundancy of each existing network point in the target city; The step of selecting at least one core grid from all target grids where new grids can be deployed, based on a preset performance threshold, the location of each existing grid point in the target city, and the predicted performance data corresponding to each target grid, includes: Based on the preset performance threshold, the existing network data of the target city, and the predicted performance data corresponding to each target grid, an incremental strategy is used to select at least one core grid from all target grids where new network points can be laid. And / or, Based on the preset performance threshold, the location and range repetition of each existing network point in the target city, and the predicted performance data corresponding to each target grid, at least one core grid that can be used to lay new network points is selected from all target grids using the stock adjustment strategy. And / or, Based on the preset performance threshold and the predicted performance data corresponding to each target grid, a full redigradation strategy is adopted to select the core grids that can be redigraded with new grid points from all target grids.

7. The method according to claim 6, characterized in that, The step of selecting at least one core grid from all target grids for new grid deployment based on a preset performance threshold, the location of each existing grid point in the target city, and the predicted performance data corresponding to each target grid, using an incremental strategy, includes: Remove all target grids within a preset radius around each existing grid point and all target grids whose predicted performance data is lower than the preset performance threshold from all target grids to obtain a grid list; The remaining target grids in the grid list are traversed sequentially. Each target grid is used as the core grid, and all target grids within a preset radius around the core grid are removed from the grid list until the grid list is empty.

8. The method according to claim 6, characterized in that, The step of selecting at least one core grid from all target grids for new grid deployment based on the preset performance threshold, the location and range repetition of each existing grid point in the target city, and the predicted performance data corresponding to each target grid, using a stock adjustment strategy, includes: Each existing network point with a range repetition degree not lower than the repetition degree threshold is designated as a network point to be replaced, and each existing network point with a range repetition degree lower than the repetition degree threshold is designated as a network point to be retained. Remove all target grids within a preset radius around each retained grid point from all target grids, as well as all target grids whose predicted performance data is lower than the preset performance threshold, to obtain a grid list; The remaining target grids in the grid list are traversed sequentially. Each target grid is used as the core grid, and all target grids within a preset radius around the core grid are removed from the grid list until the grid list is empty or the sum of the number of the core grids and the number of retained grid points is not less than a set number.

9. A network laying device, characterized in that, include: The feature extraction module is used to obtain the radiation feature vector corresponding to each target grid of the target city after the target city is processed into a grid; The radiation feature vector reflects population data, macroeconomic data, and industry data radiating outward from the center point of the target grid within a preset radius. The prediction module is used to input the radiation feature vectors corresponding to each target grid into the pre-trained performance prediction model to obtain the predicted performance data that can be achieved by laying a grid point in each target grid. The network point calculation module is used to select at least one core grid from all target grids where new network points can be laid, based on a preset performance threshold, existing network point data of the target city, and predicted performance data corresponding to each target grid.

10. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a software program, and the processor executes the software program to implement the method as described in any one of claims 1-8 when the electronic device is running.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1-8.

Images