A method, device, medium, product, and terminal equipment for predicting crowd gathering.

Abstract

This invention discloses a method, apparatus, medium, product, and terminal equipment for predicting crowd gathering, comprising: dividing the area to be predicted into zones and constructing a zone configuration table based on the divided zones; obtaining a base station configuration table for the area to be predicted based on an operator's operating parameter table; constructing a mapping relationship between zones and base stations based on the zone configuration table and the base station configuration table; obtaining operator spatiotemporal location data based on the mapping relationship and communication data between base stations and user terminals; statistically analyzing inflow and outflow data of crowd movement between zones based on the operator's spatiotemporal location data; constructing a geographic weight matrix of crowd movement between zones based on the inflow and outflow data; performing basic crowd gathering prediction on the target zone based on a preset crowd gathering prediction model, and correcting the basic prediction value based on the geographic weight matrix to obtain a real-time crowd gathering prediction value. The technical solution of this invention can ensure the real-time performance and accuracy of the prediction results.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method, apparatus, computer-readable storage medium, computer program product, and terminal equipment for predicting crowd gathering. Background Technology

[0002] With the acceleration of urbanization and the continuous increase in population size, the phenomenon of crowd gathering is becoming increasingly prominent. Large-scale events, emergencies, or holiday travel, such as major sporting events, holiday tourism, and the Spring Festival travel rush, can all lead to large gatherings of people. Crowd gatherings often cause traffic congestion and safety hazards in cities; therefore, the demand for crowd gathering prediction is constantly growing.

[0003] In today's era of rapid information technology development, an increasing amount of data can be used for crowd gathering prediction, such as public transportation data, social media data, and geographic information data. Simultaneously, the widespread application of algorithms like machine learning and deep learning has made crowd gathering prediction technology increasingly mature. Existing crowd gathering prediction methods mainly fall into two categories: one is a smoothing method based on time series data. Its basic idea is to use historical data from a past period for exponential smoothing to build a predictive model, which is then used to predict future crowd gathering trends. The other is a deep learning method based on convolutional neural networks (CNNs). Specifically, the CNN-based deep learning method can be divided into two stages: feature extraction and prediction. In the feature extraction stage, the CNN can automatically learn features from images or videos and encode them into low-dimensional vector representations. In the prediction stage, crowd gathering prediction can be performed by connecting the output of the embedding layer to a fully connected neural network.

[0004] However, the smoothing methods mentioned above based on time series data typically involve performing linear regression analysis on historical data for a specific region. The linear regression equation is then solved based on the historical data's trends. While these trends can largely reveal patterns, complex population dynamics (such as irregular travel patterns or unexpected events) can lead to significant deviations in future predictions, affecting the accuracy of the results. Convolutional neural network-based deep learning methods, while capable of handling complex population dynamics, require massive amounts of training data, multiple scenarios, and substantial hardware infrastructure. Their high computational complexity and reliance on large amounts of training data make it difficult to guarantee the real-time performance and accuracy of the predictions. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, computer-readable storage medium, computer program product, and terminal device for predicting crowd gatherings. Even under the influence of complex crowd patterns, it can more accurately predict crowd changes, and has low computational complexity, low requirements for computing resources, and low dependence on training data, thereby ensuring the real-time performance and accuracy of the prediction results.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for predicting crowd gathering, comprising: The region to be predicted is divided into regions, and a region configuration table is constructed based on the divided regions; wherein, the region configuration table includes region ID and a set of region boundary point coordinates; Obtain the base station configuration table for the area to be predicted based on the operator's operating parameters table; wherein, the base station configuration table includes base station codes and base station coordinates; The mapping relationship between areas and base stations is constructed based on the area configuration table and the base station configuration table; Based on the mapping relationship and the communication data between the base station and the user terminal, the operator's spatiotemporal location data is obtained; Based on the operator's spatiotemporal location data, statistical data on the inflow and outflow of people between areas are compiled; A geographic weight matrix of population flow between areas is constructed based on the inflow and outflow data; Based on the preset crowd gathering prediction model, a basic crowd gathering prediction is made for the target area, and the basic prediction value is corrected according to the geographic weight matrix to obtain the real-time crowd gathering prediction value.

[0007] Furthermore, the process of dividing the region to be predicted into regions and constructing a region configuration table based on the divided regions specifically includes: The area to be predicted is divided into several zones based on the boundaries of administrative divisions and a preset granularity; each zone includes at least one base station. Set a zone ID for each zone, and create a zone configuration table based on the zone ID and the set of zone boundary point coordinates for each zone.

[0008] Furthermore, the step of constructing the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table specifically includes: Based on the set of boundary point coordinates in the area configuration table and the base station coordinates in the base station configuration table, identify all base stations contained in each area; Based on all base stations contained in each area, a mapping relationship between areas and base stations is constructed; wherein, the mapping relationship is represented as the correspondence between area ID and base station code.

[0009] Furthermore, the step of obtaining operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal specifically includes: Based on the communication data between each base station and the user terminal, obtain the location information of the user terminal; Based on changes in the user terminal's location information, obtain the base station zipper entry and exit information of the user terminal; Based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice, the operator's spatiotemporal location data is obtained.

[0010] Furthermore, the step of obtaining operator spatiotemporal location data based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice specifically includes: The user terminal's entry time and exit time from the base station zipper entry and exit information are obtained from the preset time slice; When the time between entering the base station and the time between leaving the base station includes the current time slice, the base station code currently connected to the user terminal and the terminal information of the user terminal are obtained from the base station zipper entry and exit information. Based on the mapping relationship, obtain the area ID corresponding to the base station code currently connected to the user terminal, and use it as the area ID of the area where the user terminal is currently located. Based on the current time slice, the area ID of the area where the user terminal is currently located, and the terminal information of the user terminal, the operator's spatiotemporal location data is obtained.

[0011] Furthermore, the step of statistically analyzing the inflow and outflow data of population movement within a specific area based on the operator's spatiotemporal location data specifically includes: The inflow and outflow of people between areas are determined based on the operator's spatiotemporal location data. Based on the inflow and outflow diameters, statistical analysis is performed on the inflow and outflow data of the population flow between each area.

[0012] Furthermore, the step of constructing a geographic weight matrix for inter-area population flow based on the inflow and outflow data specifically includes: Based on the inflow and outflow data, the population inflow relationship and population outflow relationship between different areas are obtained; Based on the population inflow and outflow relationships between the areas, calculate the geographical weight of each area relative to its surrounding areas; Based on the geographical weights of each area and its surrounding areas, a geographical weight matrix for population flow between areas is constructed.

[0013] Furthermore, the step of calculating the geographical weight of each area relative to surrounding areas based on the population inflow and outflow relationships between the areas specifically includes: According to the formula Calculate the geographic weights of each region and its surrounding regions; where w is the geographic weight of the region to be calculated and its surrounding regions, i is the time slice index, N is the total number of time slices, and In... i Out represents the number of people flowing into the area to be calculated from the surrounding areas within the i-th time slice. i The number of people who flow out of the area to be calculated and into the surrounding areas to be calculated within the i-th time slice.

[0014] Furthermore, the step of performing basic crowd gathering prediction on the target area based on a preset crowd gathering prediction model, and correcting the basic prediction value according to the geographic weight matrix to obtain the real-time crowd gathering prediction value, specifically includes: The triple exponential smoothing method was used to make a basic prediction of population agglomeration in the target area, and the basic prediction value was obtained. The base prediction value is corrected based on the geographic weight matrix to obtain the real-time population gathering prediction value for the target area.

[0015] Furthermore, the step of correcting the basic predicted value based on the geographical weight matrix to obtain the real-time population aggregation predicted value for the target area specifically includes: According to the formula Obtain real-time crowd aggregation prediction values ​​for the target area; where y i+h x represents the predicted real-time population density of the target area at time i+h. i+h The base prediction value, j is the area index, M is the total number of surrounding areas of the target area, and w j Let ε be the geographical weight of the target area and the j-th surrounding area, where ε is a floating value.

[0016] To achieve the above objectives, embodiments of the present invention also provide a crowd gathering prediction device, comprising: The area configuration table construction module is used to divide the area to be predicted into areas and construct an area configuration table based on the divided areas; wherein, the area configuration table includes area ID and a set of area boundary point coordinates; The base station configuration table acquisition module is used to acquire the base station configuration table of the area to be predicted based on the operator's operating parameter table; wherein, the base station configuration table includes base station codes and base station coordinates; The mapping relationship construction module is used to construct the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table. The spatiotemporal location data acquisition module is used to acquire operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal. The crowd flow data statistics module is used to count the inflow and outflow data of crowd flow between areas based on the operator's spatiotemporal location data; The geographic weight matrix construction module is used to construct a geographic weight matrix of population flow between areas based on the inflow and outflow data. The crowd gathering prediction module is used to make basic crowd gathering predictions for the target area based on a preset crowd gathering prediction model, and to correct the basic prediction values ​​based on the geographic weight matrix to obtain real-time crowd gathering prediction values.

[0017] This invention also provides a computer-readable storage medium including a stored computer program, which, when running, controls the device where the computer-readable storage medium is located to execute any of the crowd gathering prediction methods described above.

[0018] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the crowd gathering prediction method described above.

[0019] This invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the crowd gathering prediction method described above when executing the computer program.

[0020] Compared with existing technologies, embodiments of the present invention provide a method, apparatus, computer-readable storage medium, computer program product, and terminal device for predicting crowd gathering. First, the area to be predicted is divided into zones, and a zone configuration table is constructed based on the divided zones. The zone configuration table includes zone IDs and a set of coordinates of zone boundary points. A base station configuration table for the area to be predicted is obtained based on the operator's operating parameters table. The base station configuration table includes base station codes and base station coordinates. A mapping relationship between zones and base stations is constructed based on the zone configuration table and the base station configuration table. Next, operator spatiotemporal location data is obtained based on the mapping relationship and communication data between base stations and user terminals. Inflow and outflow data of crowd movement between zones are statistically analyzed based on the operator's spatiotemporal location data. Then, a geographical weight matrix of crowd movement between zones is constructed based on the inflow and outflow data. Finally, a basic crowd gathering prediction is performed on the target zone based on a preset crowd gathering prediction model, and the basic prediction value is corrected based on the geographical weight matrix to obtain a real-time crowd gathering prediction value. By employing the embodiments of the present invention, even under the influence of complex population patterns, it is possible to more accurately predict population changes, and the computational complexity, computational resource requirements, and dependence on training data are all low, thereby ensuring the real-time performance and accuracy of the prediction results. Attached Figure Description

[0021] Figure 1 This is a flowchart of a crowd gathering prediction method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the population flow relationship between different areas provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a region divided according to administrative divisions, provided by an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the mapping relationship between a region and a base station according to an embodiment of the present invention; Figure 5 This is a population fluctuation curve of a region provided in one embodiment of the present invention; Figure 6 This is a multi-region population fluctuation curve provided by an embodiment of the present invention; Figure 7 This is a structural block diagram of a crowd gathering prediction device provided in an embodiment of the present invention; Figure 8 This is a structural block diagram of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0022] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that while existing crowd gathering prediction methods can meet certain data analysis needs, they have many technical shortcomings in complex environments. For example, consider linear regression on historical data: 1. Due to the influence of data quality and feature selection, linear regression models have high requirements for data quality and feature selection. If the data is noisy or the features are incomplete, the prediction results may deviate significantly. 2. It has poor fitting ability for nonlinear relationships. The linear regression model can only fit linear relationships and performs poorly in predicting nonlinear or complex relationships. 3. Sensitive to outliers: Linear regression models are easily affected by outliers. The presence of outliers may cause significant deviations in model parameters, or even render the model's predictions meaningless. 4. A large amount of historical data is required. The performance of the linear regression model is related to the amount of historical data provided. If the amount of historical data is small, the prediction results may not be accurate enough. 5. Difficulty in handling multivariate situations: If there are multiple influencing factors in the historical data of a region, the linear regression model is difficult to handle and a more complex multiple regression model may be required.

[0024] Meanwhile, as mentioned in the background section, using deep learning methods based on convolutional neural networks for prediction also has many shortcomings, mainly reflected in high computational complexity, high requirements for computing resources, high dependence on the amount of training data, resulting in poor real-time performance and accuracy, as well as low cost-effectiveness.

[0025] To address the shortcomings of existing technologies, this invention provides a method for predicting crowd aggregation. (See [link to relevant documentation]). Figure 1 The diagram shows a flowchart of a crowd gathering prediction method according to an embodiment of the present invention, the method comprising steps S11 to S17: Step S11: Divide the region to be predicted into regions and construct a region configuration table based on the divided regions; wherein, the region configuration table includes region ID and a set of region boundary point coordinates; Step S12: Obtain the base station configuration table for the area to be predicted based on the operator's operating parameter table; wherein, the base station configuration table includes base station codes and base station coordinates; Step S13: Construct a mapping relationship between areas and base stations based on the area configuration table and the base station configuration table; Step S14: Obtain operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal; Step S15: Calculate the inflow and outflow data of the population flow between the areas based on the operator's spatiotemporal location data; Step S16: Construct a geographic weight matrix of inter-area population flow based on the inflow and outflow data; Step S17: Perform basic crowd gathering prediction on the target area according to the preset crowd gathering prediction model, and correct the basic prediction value according to the geographic weight matrix to obtain the real-time crowd gathering prediction value.

[0026] Specifically, firstly, the area to be predicted is divided into several zones, and a zone configuration table is constructed based on these zones. This zone configuration table includes at least the zone ID and the set of zone boundary point coordinates for each zone. Secondly, a base station configuration table for the area to be predicted is obtained from the operator's operating parameter table. This base station configuration table includes at least the base station code and base station coordinates. Based on the zone configuration table and the base station configuration table, a mapping relationship between zones and base stations within the area to be predicted is constructed. Next, based on the mapping relationship between zones and base stations within the area to be predicted, and the relationship between base stations and user terminals (e.g., user terminals),... The system uses communication data between mobile phones to obtain the operator's spatiotemporal location data. Based on the obtained operator spatiotemporal location data, it statistically analyzes the inflow and outflow data of population movement between areas. Then, based on the statistically obtained inflow and outflow data of population movement between areas, it constructs a geographic weight matrix of population movement between areas. Finally, it performs basic population movement prediction on the target area (which can be any area within the area to be predicted) according to a pre-set population movement prediction model, obtains the corresponding basic prediction value, and corrects the obtained basic prediction value according to the constructed geographic weight matrix to finally obtain the real-time population movement prediction value of the target area.

[0027] It should be noted that a region is a geographical area formed by connecting irregular boundary points. Regions can be large or small, and their specific size is determined by the actual division results. After the regions are divided, each region can be named. Therefore, in addition to the region ID and the set of boundary point coordinates for each region, the region configuration table can also include the region name for each region, and the region name can be set according to actual needs.

[0028] For example, the set of coordinates of the boundary points of a region can be the set of latitude and longitude coordinates of the boundary points of the region. The relevant information of the region configuration table is shown in Table 1.

[0029] Table 1. Area Configuration Table

[0030] For example, the base station code can be GCI (Cell Global Identifier), the base station coordinates can be latitude and longitude coordinates, and in addition to the base station code and base station coordinates, the base station configuration table can also include other information. The relevant information in the operator's operating parameters table (the information in the base station configuration table comes from the operator's operating parameters table) is shown in Table 2.

[0031] Table 2 Operator Operating Parameters

[0032] It should be noted that spatial data is widely present in many fields such as geography, economics, environmental science, ecology, and meteorology. Everything is spatially related, and the closer things are, the greater the spatial correlation. Therefore, unlike traditional cross-sectional data, the spatial correlation of spatial data will lead to spatial non-stationarity (spatial heterogeneity) of the regression relationship. In order to solve the spatial non-stationarity of spatial data, this invention constructs a geographical weight matrix of population flow between areas to weight geographical areas, so as to explore the correlation factors of a certain area under the spatiotemporal changes of the surrounding population.

[0033] For example, see Figure 2 The diagram shown is a schematic diagram of inter-area population flow relationship provided by an embodiment of the present invention. Assuming that the area to be predicted is divided into 9 areas, namely area a and areas 1 to 8, and area a is taken as the target area, the population gathering prediction of target area a is closely related to the population inflow and outflow between its adjacent areas and even other non-adjacent areas. Therefore, the inter-area population flow analysis can be carried out by combining the surrounding areas (areas 1 to 8) of target area a, and a geographical weight matrix of inter-area population flow can be constructed to analyze the trend of population inflow and outflow, and then analyze the future population gathering trend in target area a.

[0034] The crowd gathering prediction method provided in this invention divides the area into zones and maps base stations. Based on operator-specific spatiotemporal location data, it performs geographically weighted modeling of each zone and its neighboring zones to explore the correlation between a zone and the spatiotemporal changes of the surrounding population. Then, it combines linear regression of the most recent sample data of the population in that zone to obtain the corrected future crowd gathering trend, thereby more accurately predicting crowd changes. It can be seen that by adopting this invention, even under the influence of complex crowd patterns, it is possible to more accurately predict crowd changes. Moreover, the computational complexity, the requirements for computing resources, and the dependence on the amount of training data are all low, thus ensuring the real-time performance and accuracy of the prediction results.

[0035] In one optional embodiment, the step of dividing the region to be predicted into regions and constructing a region configuration table based on the divided regions specifically includes: The area to be predicted is divided into several zones based on the boundaries of administrative divisions and a preset granularity; each zone includes at least one base station. Set a zone ID for each zone, and create a zone configuration table based on the zone ID and the set of zone boundary point coordinates for each zone.

[0036] Specifically, in conjunction with the above embodiments, when dividing the area to be predicted into zones and constructing a zone configuration table based on the divided zones, the area to be predicted can first be divided into several zones according to the boundaries of administrative divisions and a preset granularity. Then, a zone ID (the zone ID is unique and can be automatically generated) can be set for each zone. Based on the zone ID of each zone and the coordinate set of zone boundary points (i.e., the boundary points of administrative divisions), a zone configuration table can be established.

[0037] It should be noted that each area includes at least one base station (e.g., 1 to 100,000) to facilitate the subsequent establishment of a mapping relationship between areas and base stations.

[0038] For example, see Figure 3 The diagram shown is a schematic diagram of a district divided according to administrative divisions according to an embodiment of the present invention. That is, the district can be divided according to the boundaries of administrative divisions, with the smallest granularity being the business district. Depending on the attributes of the business district, if it is a densely populated area, the district can be divided with a granularity of 100×100 meters. If it is a sparsely populated area, the district should be divided with a minimum granularity of 1000×1000 meters.

[0039] Understandably, if districts are divided according to administrative boundaries, the boundaries of these districts are the natural boundaries of the administrative divisions, and generally will not be as... Figure 2 The rectangular boundary shown, therefore, Figure 2 The illustration shown is merely an example.

[0040] In one optional embodiment, the step of constructing the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table specifically includes: Based on the set of boundary point coordinates in the area configuration table and the base station coordinates in the base station configuration table, identify all base stations contained in each area; Based on all base stations contained in each area, a mapping relationship between areas and base stations is constructed; wherein, the mapping relationship is represented as the correspondence between area ID and base station code.

[0041] Specifically, in conjunction with the above embodiments, when constructing the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table, we can first identify all the base stations contained in each area (i.e., determine which base stations are specifically contained in each area) based on the set of area boundary point coordinates in the area configuration table and the base station coordinates in the base station configuration table. Then, based on all the base stations contained in each area, we can construct the mapping relationship between areas and base stations. This mapping relationship can be represented as the correspondence between area ID and base station code.

[0042] For example, the mapping relationship between the area and the base station is shown in Table 3.

[0043] Table 3 Mapping Relationship between Area and Base Station

[0044] For example, the set of boundary point coordinates for a region can be obtained based on the `point` field in Table 1, and the base station coordinates can be obtained based on the `Latitude` and `Longitude` fields in Table 2. (See [link to Table 2]). Figure 4 The diagram shown is a schematic diagram of the mapping relationship between a region and a base station provided in an embodiment of the present invention. Assume that the coordinate set of the boundary points of region 1 is {A(x1,y1), B(x2,y2), C(x3,y3), D(x4,y4), E(x5,y5)}, the base station coordinate of base station 1 is (x6,y6), and the base station coordinate of base station 2 is (x7,y7). After coordinate judgment, it can be seen that base station 1 is inside region 1, while base station 2 is outside region 1. Therefore, there is a mapping relationship between region 1 and base station 1, and the mapping relationship between region ID of region 1 and base station code of base station 1 can be constructed.

[0045] Understandably, if districts are divided according to administrative boundaries, the boundaries of these districts are the natural boundaries of the administrative divisions, and generally will not be as... Figure 4 The pentagonal boundary (or other polygonal boundary) shown, therefore, Figure 4 The illustration shown is merely an example; in practice, the boundary points of region 1 can be described using the WKT (Well-Known Text) standard format: POLYGON(( <x1> <y1> ， <x2> <y2>Similarly, Figure 4 The coordinates shown are in the Cartesian coordinate system xOy, and are only an example. In practice, latitude and longitude coordinates can be used.

[0046] It should be noted that the embodiments of the present invention are based on the operator's base station operating parameter table. By constructing the mapping relationship between the area and the base station location through the positional relationship between the point (base station) and the polygon (area), compared with other prediction methods that use the latitude and longitude of the population for location prediction, the embodiments of the present invention use the base station location as the positioning data, which can obtain higher population coverage, more stable location data quality, more efficient spatial aggregation and real-time performance, and lower business implementation costs (no need for additional user terminal or hardware modification, and no need to rely on third-party positioning).

[0047] In one optional embodiment, obtaining operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal specifically includes: Based on the communication data between each base station and the user terminal, obtain the location information of the user terminal; Based on changes in the user terminal's location information, obtain the base station zipper entry and exit information of the user terminal; Based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice, the operator's spatiotemporal location data is obtained.

[0048] Specifically, in conjunction with the above embodiments, when obtaining operator spatiotemporal location data based on the mapping relationship between the area and the base station and the communication data between the base station and the user terminal, the location information of the user terminal can be obtained first based on the communication data between each base station and the user terminal, and the base station zipper entry and exit information of the user terminal can be obtained based on the changes in the location information of the user terminal. Then, based on the mapping relationship between the area and the base station, the base station zipper entry and exit information of the user terminal and the pre-set time slice, the operator spatiotemporal location data can be obtained.

[0049] It should be noted that user terminals (such as user mobile phones) will connect to the nearest base station to the user in real time. If it can be determined which base station the user mobile phone is connected to, the location information of the user mobile phone can be known by obtaining the latitude and longitude of the base station. Furthermore, as the user moves, the user mobile phone will continuously switch to connect to different base stations, and the location information of the user mobile phone will also change continuously, which will form the base station zipper entry and exit information of the user mobile phone.

[0050] For example, the base station zipper entry and exit information of the user terminal is shown in Table 4.

[0051] Table 4 Base Station Zipper Access Information Table

[0052] In one optional embodiment, obtaining operator spatiotemporal location data based on the mapping relationship, the base station zipper entry and exit information, and a preset time slice specifically includes: The user terminal's entry time and exit time from the base station zipper entry and exit information are obtained from the preset time slice; When the time between entering the base station and the time between leaving the base station includes the current time slice, the base station code currently connected to the user terminal and the terminal information of the user terminal are obtained from the base station zipper entry and exit information. Based on the mapping relationship, obtain the area ID corresponding to the base station code currently connected to the user terminal, and use it as the area ID of the area where the user terminal is currently located. Based on the current time slice, the area ID of the area where the user terminal is currently located, and the terminal information of the user terminal, the operator's spatiotemporal location data is obtained.

[0053] Specifically, in conjunction with the above embodiments, when obtaining operator spatiotemporal location data based on the mapping relationship between the area and the base station, the base station zipper entry and exit information of the user terminal, and the pre-set time slice, the entry time and exit time of the user terminal can be obtained from the base station zipper entry and exit information according to the pre-set time slice (e.g., 5 minutes). When the entry time and exit time of the user terminal include the current time slice, the base station code of the base station currently connected to the user terminal and the terminal information of the user terminal (e.g., the user's mobile phone number) can be obtained from the base station zipper entry and exit information. Based on the mapping relationship between the area and the base station, the area ID corresponding to the base station code of the base station currently connected to the user terminal can be obtained as the area ID of the area where the user terminal is currently located. Finally, based on the current time slice, the area ID of the area where the user terminal is currently located, and the terminal information of the user terminal, the operator spatiotemporal location data can be obtained.

[0054] It should be noted that, since each user's travel patterns are different, for ease of calculation, a 5-minute location slice method can be used to record the area where each user is currently located every 5 minutes, thus obtaining the operator's spatiotemporal location data, as shown in Table 5.

[0055] Table 5. Spatiotemporal Location Data of Operators

[0056] The slicing rules are as follows: A slice is defined as 5 minutes. Every 5 minutes at the top of the hour (time), the procedureStartTime and procedureEndTime fields from Table 4 are retrieved. If the currently retrieved procedureStartTime <= time <= procedureEndTime, i.e., [procedureStartTime, procedureEndTime] includes the current 5-minute time, it indicates that the user's mobile phone is in a connected state during this period. Then, the CGI and msisdn fields from Table 4 are retrieved. The CGI field from Table 4 is matched with the CGI field from Table 3 to obtain the corresponding area_id, which is then filled into the area_id field in Table 5. Finally, the msisdn field from Table 4 is filled into the msisdn field in Table 5, thus obtaining the operator's spatiotemporal location data shown in Table 5.

[0057] It should be noted that the purpose of determining whether procedureStartTime<=time<=procedureEndTime is to further form a 5-minute regional snapshot based on the base station zipper entry and exit information shown in Table 4, i.e., the operator spatiotemporal location data shown in Table 5. By retrieving data from the base station zipper entry and exit information table according to a 5-minute time granularity, operator spatiotemporal location data can be obtained for each 5-minute interval based on the procedureStartTime and procedureEndTime fields.

[0058] In one optional embodiment, the step of statistically analyzing the inflow and outflow data of population movement within a segment based on the operator's spatiotemporal location data specifically includes: The inflow and outflow of people between areas are determined based on the operator's spatiotemporal location data. Based on the inflow and outflow diameters, statistical analysis is performed on the inflow and outflow data of the population flow between each area.

[0059] Specifically, in conjunction with the above embodiments, when calculating the inflow and outflow data of population flow between areas based on the operator's spatiotemporal location data, the inflow and outflow paths of population flow between areas can be determined first based on the operator's spatiotemporal location data. Then, based on the determined inflow and outflow paths, the inflow and outflow data of population flow between areas corresponding to each area can be calculated.

[0060] For example, in combination Figure 2 As shown, assuming the inflow direction is: if a user with a 5-minute granularity is in target area a, and the previous user with a 5-minute granularity is in area 1, it means that target area a has an inflow to area 1; assuming the outflow direction is: if a user with a 5-minute granularity is in target area a, and the next user with a 5-minute granularity is in area 1, it means that target area a has an outflow to area 1. Accordingly, by summarizing the inflow and outflow of each user in the area, the overall inflow and outflow data of the area can be obtained, as shown in Table 6.

[0061] Table 6. Inflow and outflow data of population movement between areas

[0062] It should be noted that the embodiments of the present invention are based on the spatiotemporal location data of the operator to count the inflow and outflow of people in the area, which has a larger data model coverage and higher accuracy than other prediction algorithms.

[0063] In one optional embodiment, constructing a geographic weight matrix for inter-area population flow based on the inflow and outflow data specifically includes: Based on the inflow and outflow data, the population inflow relationship and population outflow relationship between different areas are obtained; Based on the population inflow and outflow relationships between the areas, calculate the geographical weight of each area relative to its surrounding areas; Based on the geographical weights of each area and its surrounding areas, a geographical weight matrix for population flow between areas is constructed.

[0064] Specifically, in conjunction with the above embodiments, when constructing the geographic weight matrix of inter-regional population flow based on the inflow and outflow data of inter-regional population flow, the inflow relationship and outflow relationship of inter-regional population flow can be obtained first. Then, based on the obtained inflow and outflow relationships of inter-regional population flow, the geographic weight of each region relative to each of its surrounding regions can be calculated. Finally, based on the geographic weight of each region and each of its surrounding regions, the geographic weight matrix of inter-regional population flow corresponding to the entire area to be predicted can be constructed.

[0065] In one optional embodiment, calculating the geographical weight of each area relative to surrounding areas based on the population inflow and outflow relationships between the areas specifically includes: According to the formula Calculate the geographic weights of each region and its surrounding regions; where w is the geographic weight of the region to be calculated and its surrounding regions, i is the time slice index, N is the total number of time slices, and In... i Out represents the number of people flowing into the area to be calculated from the surrounding areas within the i-th time slice. i The number of people who flow out of the area to be calculated and into the surrounding areas to be calculated within the i-th time slice.

[0066] Specifically, in conjunction with the above embodiments, the formula for calculating the geographical weight of each area relative to any of its surrounding areas is as follows: w represents the geographic weight of the area to be calculated and a surrounding area (as the surrounding area to be calculated), i represents the time slice index, N represents the total number of time slices, and In i Out represents the number of people flowing into the area to be calculated from the surrounding areas within the i-th time slice. i This represents the number of people who flow out of the area to be calculated and into the surrounding areas within the i-th time slice.

[0067] It should be noted that if w is positive, it indicates inflow; if w is negative, it indicates outflow. w is a dynamic result to ensure that the data can reflect recent population gathering trends.

[0068] For example, with Figure 2 Taking the example shown in Table 6, we can obtain the inter-area population inflow and outflow relationships for each area. The principle for obtaining the inter-area population inflow relationship is as follows: take the data with flow_type=1 in the table, and construct an inter-area population inflow relationship table based on the population flow of all areas, as shown in Table 7. The data in Table 7 is the count field in Table 6 (for example, the number of people flowing from target area a into area 1 is 500, and the number of people flowing from target area a into area 2 is 600). The principle for obtaining the inter-area population outflow relationship is as follows: take the data with flow_type=2 in Table 6, and construct an inter-area population outflow relationship table based on the population flow of all areas, as shown in Table 8. The data in Table 8 is the count field in Table 6 (for example, the number of people flowing out of target area a into area 1 is 130, and the number of people flowing out of target area a into area 2 is 80).

[0069] Table 7. Relationship of Population Inflow Between Areas

[0070] Table 8. Relationship between Population Outflows from Different Areas

[0071] It should be noted that the time slices corresponding to Tables 7 and 8 should be the same as the time field in Table 6, and in Tables 7 and 8, positive numbers represent inflows and negative numbers represent outflows.

[0072] Furthermore, the population migration weights (i.e., geographical weights) between geographical slices can be obtained from the differences between population inflows and outflows in Tables 7 and 8. To obtain more accurate weight values, the average of the inflow and outflow differences from multiple historical time slices can be used to calculate... Figure 2 Taking the geographic weights of target area a and area 1 as an example, the calculation formula is as follows: w1 represents the geographic weight of target region a and region 1, i represents the time slice index, and N represents the total number of time slices (it is recommended to use one hour of data as the statistical base). i Out represents the number of people flowing from region 1 into target region a within the i-th time slice. i This represents the number of people who flow out of target region a and into region 1 within the i-th time slice.

[0073] Furthermore, after calculating the geographical weights of each area and each of its surrounding areas, a geographical weight matrix W for population flow between areas can be constructed, as follows: .

[0074] It should be noted that, through the weighting of geographical location, the embodiments of the present invention can realize the linkage of crowd gathering prediction in the overall geographic space, thereby improving the accuracy of the prediction results.

[0075] In one optional embodiment, the step of performing basic crowd gathering prediction on the target area based on a preset crowd gathering prediction model, and correcting the basic prediction value according to the geographic weight matrix to obtain a real-time crowd gathering prediction value, specifically includes: The triple exponential smoothing method was used to make a basic prediction of population agglomeration in the target area, and the basic prediction value was obtained. The base prediction value is corrected based on the geographic weight matrix to obtain the real-time population gathering prediction value for the target area.

[0076] Specifically, in conjunction with the above embodiments, when performing basic crowd gathering prediction on the target area based on a preset crowd gathering prediction model, and correcting the basic prediction value based on the geographic weight matrix to obtain the real-time crowd gathering prediction value, the cubic exponential smoothing method can be used as the crowd gathering prediction model first. The basic crowd gathering prediction on the target area can be performed based on the cubic exponential smoothing method to obtain the corresponding basic prediction value. Then, the obtained basic prediction value can be corrected based on the constructed geographic weight matrix of crowd flow between areas to finally obtain the real-time crowd gathering prediction value of the target area.

[0077] In one optional embodiment, the step of correcting the basic prediction value based on the geographic weight matrix to obtain the real-time crowd aggregation prediction value for the target area specifically includes: According to the formula Obtain real-time crowd aggregation prediction values ​​for the target area; where y i+h x represents the predicted real-time population density of the target area at time i+h. i+h The base prediction value, j is the area index, M is the total number of surrounding areas of the target area, and w j Let ε be the geographical weight of the target area and the j-th surrounding area, where ε is a floating value.

[0078] Specifically, in conjunction with the above embodiments, when correcting the basic predicted value based on the geographic weight matrix, the correction can be performed according to the following formula: y i+h x represents the predicted real-time population density in the target area at time i+h. i+h This represents the basic predicted population density of the target area at time i+h, obtained using the triple exponential smoothing method. j represents the area index, M represents the total number of surrounding areas of the target area, and w represents the population density of the target area. j ε represents the geographical weight of the target area and its j-th surrounding area, where ε is a floating value.

[0079] It should be noted that time series analysis is a statistical analysis method that establishes mathematical models based on time series data and the interdependencies between series through curve fitting and parameter estimation. This method is suitable for non-stationary series with linear trends and fixed periods, such as… Figure 5 The figure shown is a population fluctuation curve of a region provided by an embodiment of the present invention. The horizontal axis represents time, and the vertical axis represents the number of people. Figure 5 This shows the changing trend of crowd gathering in a certain area within a day, obtained using time series analysis methods, with obvious crowd gathering phenomena during the morning and evening peak hours.

[0080] It should be noted that population gathering in a given area is related to the historical population of that area. Population movement, within a very short time window, will not experience a sudden surge or drop (considering factors such as transportation capacity and travel modes). Therefore, time series (dynamic series) analysis methods can be used to solve this problem in the short term. This method is applicable to non-stationary series containing linear trends and periodic fluctuations. Then, exponential smoothing is used to continuously adapt the model parameters to the changes in the non-stationary series, providing short-term forecasts of future trends. Considering the seasonal patterns of population travel in the area, future population gathering can be predicted by combining "current level + trend change + periodic pattern." Therefore, this embodiment of the invention can use triple exponential smoothing to perform basic predictions of population gathering, as shown in the following formula: , Therefore, the prediction formula for the basic forecast value is as follows: , Where i represents the current time step (moment), such as the i-th hour or the i-th day. In crowd gathering prediction, it represents a specific point in time being processed (such as the 3rd day or the 15th hour), and is an index in the time series; h represents the prediction time step (span), that is, how many time units in the future the crowd size is to be predicted starting from the current time i. For example, h=2 means predicting the crowd size in the 2nd time unit after the current time i (i.e., the crowd size at the future time i+h); k represents the cycle length of crowd gathering. Since the travel patterns of people in a region are seasonal (such as weekdays / weekends, morning rush hour / evening rush hour), k refers to the length of this cycle; s i t represents the smoothed value (horizontal component) at time i, which is a smoothed estimate of the "baseline level" of the crowd size at time i, reflecting the core size of the sequence after removing trends and periods; i This represents the trend component at time i, reflecting the changing trend (increase / decrease) of the crowd size at time i, such as the magnitude of the trend of "an increase of 50 people per hour"; p i The periodic component at time i refers to the "periodic" part, reflecting the periodic fluctuations in crowd gathering. For example, the number of people during the morning rush hour will regularly increase and then regularly decrease after the evening rush hour; x i This represents the actual observed value at time i, i.e., the actual number of people gathered at time i through equipment or statistics, which is the original input data of the entire time series; the values ​​of α, β, and γ are all between [0, 1], and multiple trials can be conducted to achieve the best results. The selection of the initial values ​​of s, t, and p does not have a particularly large impact on the overall algorithm. The usual values ​​are s0=x0, t0=x1-x0, and p=0 when accumulating (accumulation can be used in the embodiments of this invention).

[0081] It should be noted that while triple exponential smoothing combined with time series data can predict future population gathering, it cannot reflect the impact of population changes in surrounding areas. Therefore, it needs to be corrected based on the geographic weight matrix. The prediction formula for real-time population gathering is: , Among them, y i+h x represents the predicted real-time population density in the target area at time i+h. i+h This represents the base forecast value obtained using the triple exponential smoothing method, where j represents the area index, M represents the total number of surrounding areas of the target area, and w j This represents the geographical weight of the target area and its j-th surrounding area, where ε is a floating value (e.g., a floating value under emergency circumstances).

[0082] Understandably, the principle for determining the floating value is: under normal circumstances, ε=0; in extreme cases, where the inflow and outflow between two areas cannot be completed, ε=w, where w is the geographical weight between the two areas.

[0083] It should be noted that by combining the base forecast value obtained through triple exponential smoothing of the time series data with the geographical weights of surrounding areas and adjustments for fluctuations, the real-time population aggregation forecast value for a certain area at a future time point can be obtained. Similarly, population aggregation trend maps for multiple areas can be obtained, such as... Figure 6 The figure shown is a multi-area population fluctuation curve provided by an embodiment of the present invention. The horizontal axis represents time, and the vertical axis represents the number of people. Figure 6 This shows the changing trends of crowd gathering in Area 1, Area 2 and Target Area a throughout the day. In addition to the morning and evening peak hours, crowd gathering also occurs at other times.

[0084] It should be noted that since people's travel is greatly related to their subjective will, irregular population movement cannot be ruled out. This is an extreme case and will not be treated specially here.

[0085] It should be noted that with the continuous development of information technology and the increasing sophistication of 5G network construction, mobile phone user spatiotemporal location data from operators plays a significant role in many fields such as transportation, emergency command, public safety, and public services. In particular, the prediction and analysis of crowd gathering has immense commercial value in industry applications. (1) Emergency response: By predicting the real-time population gathering in the area, densely populated areas can be identified in advance, and emergency responses can be made in a timely manner to reduce casualties and property losses; (2) Traffic flow management: By predicting the location and time of crowd gathering, traffic routes can be optimized and the flow of people can be managed, thereby reducing urban traffic congestion and crowding. (3) Riot control: By predicting the scale and location of crowd gatherings, security departments can take more targeted anti-terrorism and anti-riot control measures; (4) Business opportunities: Traffic congestion sites, such as stations, airports, and subways, are usually places with high traffic flow. These areas have high commercial value. Businesses can rent shops in these places to open stores, fast food restaurants, convenience stores, etc., to attract customers and increase sales revenue. (5) Transportation services: Traffic management sites can provide transportation services such as taxis, buses, and subways. Businesses can cooperate with transportation service providers to provide information on tourist attractions, restaurants, hotels, etc. in these sites to provide convenient services to customers.

[0086] This invention also provides a crowd gathering prediction device for implementing the crowd gathering prediction method described in any of the above embodiments. See [link to relevant documentation]. Figure 7 The diagram shown is a structural block diagram of a crowd gathering prediction device according to an embodiment of the present invention. The device includes: The area configuration table construction module 11 is used to divide the area to be predicted into areas and construct an area configuration table based on the divided areas; wherein, the area configuration table includes area ID and a set of area boundary point coordinates; The base station configuration table acquisition module 12 is used to acquire the base station configuration table of the area to be predicted based on the operator's operating parameter table; wherein, the base station configuration table includes base station codes and base station coordinates; The mapping relationship construction module 13 is used to construct a mapping relationship between areas and base stations based on the area configuration table and the base station configuration table. The spatiotemporal location data acquisition module 14 is used to acquire operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal. The crowd flow data statistics module 15 is used to count the inflow and outflow data of crowd flow between areas based on the operator's spatiotemporal location data; The geographic weight matrix construction module 16 is used to construct a geographic weight matrix of population flow between areas based on the inflow and outflow data. The crowd gathering prediction module 17 is used to make a basic prediction of crowd gathering in the target area according to the preset crowd gathering prediction model, and to correct the basic prediction value according to the geographic weight matrix to obtain the real-time crowd gathering prediction value.

[0087] Preferably, the area configuration table construction module 11 specifically includes: The area division unit is used to divide the area to be predicted into several areas according to the boundaries of administrative divisions and preset granularity; each area includes at least one base station. The area configuration table construction unit is used to set an area ID for each area and to build an area configuration table based on the area ID and the set of area boundary point coordinates for each area.

[0088] Preferably, the mapping relationship construction module 13 specifically includes: The base station identification unit is used to identify all base stations contained in each area based on the set of area boundary point coordinates in the area configuration table and the base station coordinates in the base station configuration table. The mapping relationship construction unit is used to construct a mapping relationship between areas and base stations based on all base stations contained in each area; wherein, the mapping relationship is represented as the correspondence between area ID and base station code.

[0089] Preferably, the spatiotemporal location data acquisition module 14 specifically includes: The location information acquisition unit is used to acquire the location information of the user terminal based on the communication data between each base station and the user terminal; The base station zipper information acquisition unit is used to acquire the base station zipper entry and exit information of the user terminal based on the changes in the location information of the user terminal. The spatiotemporal location data acquisition unit is used to acquire operator spatiotemporal location data based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice.

[0090] Preferably, the spatiotemporal location data acquisition unit is specifically used for: The user terminal's entry time and exit time from the base station zipper entry and exit information are obtained from the preset time slice; When the time between entering the base station and the time between leaving the base station includes the current time slice, the base station code currently connected to the user terminal and the terminal information of the user terminal are obtained from the base station zipper entry and exit information. Based on the mapping relationship, obtain the area ID corresponding to the base station code currently connected to the user terminal, and use it as the area ID of the area where the user terminal is currently located. Based on the current time slice, the area ID of the area where the user terminal is currently located, and the terminal information of the user terminal, the operator's spatiotemporal location data is obtained.

[0091] Preferably, the crowd flow data statistics module 15 specifically includes: The crowd flow caliber determination unit is used to determine the inflow and outflow calibers of crowd flow between areas based on the operator's spatiotemporal location data; The crowd flow data statistics unit is used to count the inflow and outflow data of crowd flow between each area based on the inflow inlet diameter and the outflow outlet diameter.

[0092] Preferably, the geographic weight matrix construction module 16 specifically includes: The crowd flow relationship acquisition unit is used to acquire the crowd inflow relationship and the crowd outflow relationship between areas based on the inflow and outflow data. The area geographic weight calculation unit is used to calculate the geographic weight of each area and its surrounding areas based on the population inflow relationship and population outflow relationship between the areas. The geographic weight matrix construction unit is used to construct a geographic weight matrix of population flow between areas based on the geographic weights of each area and its surrounding areas.

[0093] Preferably, the area geographic weight calculation unit is specifically used for: According to the formula Calculate the geographic weights of each region and its surrounding regions; where w is the geographic weight of the region to be calculated and its surrounding regions, i is the time slice index, N is the total number of time slices, and In... i Out represents the number of people flowing into the area to be calculated from the surrounding areas within the i-th time slice. i The number of people who flow out of the area to be calculated and into the surrounding areas to be calculated within the i-th time slice.

[0094] Preferably, the crowd gathering prediction module 17 specifically includes: The crowd gathering basic prediction unit is used to perform basic crowd gathering prediction on the target area using the triple exponential smoothing method to obtain basic prediction values. The real-time crowd gathering prediction unit is used to correct the basic prediction value based on the geographic weight matrix to obtain the real-time crowd gathering prediction value of the target area.

[0095] Preferably, the real-time crowd gathering prediction unit is specifically used for: According to the formula Obtain real-time crowd aggregation prediction values ​​for the target area; where y i+h x represents the predicted real-time population density of the target area at time i+h. i+h The base prediction value, j is the area index, M is the total number of surrounding areas of the target area, and w j Let ε be the geographical weight of the target area and the j-th surrounding area, where ε is a floating value.

[0096] It should be noted that the crowd gathering prediction device provided in this embodiment of the invention can realize all the processes of the crowd gathering prediction method described in any of the above embodiments. The functions and technical effects of each module and unit in the device are the same as the functions and technical effects of the crowd gathering prediction method described in the above embodiments, and will not be repeated here.

[0097] This invention also provides a computer-readable storage medium including a stored computer program, which, when running, controls the device containing the computer-readable storage medium to execute the crowd gathering prediction method described in any of the above embodiments.

[0098] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the crowd gathering prediction method described in any of the above embodiments.

[0099] This invention also provides a terminal device, see [link to relevant documentation]. Figure 8 The diagram shown is a structural block diagram of a terminal device provided in an embodiment of the present invention. The terminal device includes a processor 10, a memory 20, and a computer program stored in the memory 20 and configured to be executed by the processor 10. When the processor 10 executes the computer program, it implements the crowd gathering prediction method described in any of the above embodiments.

[0100] Preferably, the computer program can be divided into one or more modules / units (such as computer program 1, computer program 2, ...), and the one or more modules / units are stored in the memory 20 and executed by the processor 10 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0101] The processor 10 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor 10 may be any conventional processor. The processor 10 is the control center of the terminal device, connecting various parts of the terminal device through various interfaces and lines.

[0102] The memory 20 mainly includes a program storage area and a data storage area. The program storage area can store the operating system, applications required for at least one function, etc., while the data storage area can store related data, etc. Furthermore, the memory 20 can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard drive, a smart media card (SMC), a secure digital card (SD), and a flash card, or other volatile solid-state storage devices.

[0103] It should be noted that the aforementioned terminal devices may include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 8 The structural block diagram shown is merely a structural example of the terminal device described above and does not constitute a limitation on the structure of the terminal device. The terminal device may include more or fewer components than shown, or combine certain components, or use different components.

[0104] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. < / x2> < / y1> < / x1>

Claims

1. A method for predicting crowd aggregation, characterized in that, include: The region to be predicted is divided into regions, and a region configuration table is constructed based on the divided regions; wherein, the region configuration table includes region ID and a set of region boundary point coordinates; Obtain the base station configuration table for the area to be predicted based on the operator's operating parameters table; wherein, the base station configuration table includes base station codes and base station coordinates; The mapping relationship between areas and base stations is constructed based on the area configuration table and the base station configuration table; Based on the mapping relationship and the communication data between the base station and the user terminal, the operator's spatiotemporal location data is obtained; Based on the operator's spatiotemporal location data, statistical data on the inflow and outflow of people between areas are compiled; A geographic weight matrix of population flow between areas is constructed based on the inflow and outflow data; Based on the preset crowd gathering prediction model, a basic crowd gathering prediction is made for the target area, and the basic prediction value is corrected according to the geographic weight matrix to obtain the real-time crowd gathering prediction value.

2. The crowd aggregation prediction method as described in claim 1, characterized in that, The process of dividing the region to be predicted into regions and constructing a region configuration table based on the divided regions specifically includes: The area to be predicted is divided into several zones based on the boundaries of administrative divisions and a preset granularity; each zone includes at least one base station. Set a zone ID for each zone, and create a zone configuration table based on the zone ID and the set of zone boundary point coordinates for each zone.

3. The crowd gathering prediction method as described in claim 1, characterized in that, The step of constructing the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table specifically includes: Based on the set of boundary point coordinates in the area configuration table and the base station coordinates in the base station configuration table, identify all base stations contained in each area; Based on all base stations contained in each area, a mapping relationship between areas and base stations is constructed; wherein, the mapping relationship is represented as the correspondence between area ID and base station code.

4. The crowd aggregation prediction method as described in claim 1, characterized in that, The step of obtaining operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal specifically includes: Based on the communication data between each base station and the user terminal, obtain the location information of the user terminal; Based on changes in the user terminal's location information, obtain the base station zipper entry and exit information of the user terminal; Based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice, the operator's spatiotemporal location data is obtained.

5. The crowd aggregation prediction method as described in claim 4, characterized in that, The step of obtaining operator spatiotemporal location data based on the mapping relationship, the base station zipper entry and exit information, and the preset time slice specifically includes: The user terminal's entry time and exit time from the base station zipper entry and exit information are obtained from the preset time slice; When the time between entering the base station and the time between leaving the base station includes the current time slice, the base station code currently connected to the user terminal and the terminal information of the user terminal are obtained from the base station zipper entry and exit information. Based on the mapping relationship, obtain the area ID corresponding to the base station code currently connected to the user terminal, and use it as the area ID of the area where the user terminal is currently located. Based on the current time slice, the area ID of the area where the user terminal is currently located, and the terminal information of the user terminal, the operator's spatiotemporal location data is obtained.

6. The crowd aggregation prediction method as described in claim 1, characterized in that, The step of statistically analyzing the inflow and outflow data of population movement within a specific area based on the operator's spatiotemporal location data specifically includes: The inflow and outflow of people between areas are determined based on the operator's spatiotemporal location data. Based on the inflow and outflow diameters, statistical analysis is performed on the inflow and outflow data of the population flow between each area.

7. The crowd aggregation prediction method as described in claim 1, characterized in that, The construction of a geographic weight matrix for inter-regional population flow based on the inflow and outflow data specifically includes: Based on the inflow and outflow data, the population inflow relationship and population outflow relationship between different areas are obtained; Based on the population inflow and outflow relationships between the areas, calculate the geographical weight of each area relative to its surrounding areas; Based on the geographical weights of each area and its surrounding areas, a geographical weight matrix for population flow between areas is constructed.

8. The crowd gathering prediction method as described in claim 7, characterized in that, The step of calculating the geographical weight of each area relative to surrounding areas based on the population inflow and outflow relationships between the areas specifically includes: According to the formula Calculate the geographic weights of each region and its surrounding regions; where w is the geographic weight of the region to be calculated and its surrounding regions, i is the time slice index, N is the total number of time slices, and In... i Out represents the number of people flowing into the area to be calculated from the surrounding areas within the i-th time slice. i The number of people who flow out of the area to be calculated and into the surrounding areas to be calculated within the i-th time slice.

9. The crowd aggregation prediction method as described in claim 1, characterized in that, The step of performing basic crowd gathering prediction on the target area based on a preset crowd gathering prediction model, and then correcting the basic prediction value based on the geographic weight matrix to obtain the real-time crowd gathering prediction value, specifically includes: The triple exponential smoothing method was used to make a basic prediction of population agglomeration in the target area, and the basic prediction value was obtained. The base prediction value is corrected based on the geographic weight matrix to obtain the real-time population gathering prediction value for the target area.

10. The crowd aggregation prediction method as described in claim 9, characterized in that, The step of correcting the basic prediction value based on the geographic weight matrix to obtain the real-time population aggregation prediction value for the target area specifically includes: According to the formula Obtain real-time crowd aggregation prediction values ​​for the target area; where y i+h x represents the predicted real-time population density of the target area at time i+h. i+h The base prediction value, j is the area index, M is the total number of surrounding areas of the target area, and w j Let ε be the geographical weight of the target area and the j-th surrounding area, where ε is a floating value.

11. A crowd gathering prediction device, characterized in that, include: The area configuration table construction module is used to divide the area to be predicted into areas and construct an area configuration table based on the divided areas; wherein, the area configuration table includes area ID and a set of area boundary point coordinates; The base station configuration table acquisition module is used to acquire the base station configuration table of the area to be predicted based on the operator's operating parameter table; wherein, the base station configuration table includes base station codes and base station coordinates; The mapping relationship construction module is used to construct the mapping relationship between areas and base stations based on the area configuration table and the base station configuration table. The spatiotemporal location data acquisition module is used to acquire operator spatiotemporal location data based on the mapping relationship and communication data between the base station and the user terminal. The crowd flow data statistics module is used to count the inflow and outflow data of crowd flow between areas based on the operator's spatiotemporal location data; The geographic weight matrix construction module is used to construct a geographic weight matrix of population flow between areas based on the inflow and outflow data. The crowd gathering prediction module is used to make basic crowd gathering predictions for the target area based on a preset crowd gathering prediction model, and to correct the basic prediction values ​​based on the geographic weight matrix to obtain real-time crowd gathering prediction values.

12. A computer-readable storage medium, characterized in that, The device includes a stored computer program that, when run, controls the device containing the computer-readable storage medium to perform the crowd gathering prediction method as described in any one of claims 1 to 10.

13. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the crowd gathering prediction method as described in any one of claims 1 to 10.

14. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the crowd gathering prediction method as described in any one of claims 1 to 10.

Images