Telephone traffic demand prediction method and system

Abstract

The invention provides a telephone traffic demand prediction method and system. The method comprises the steps that firstly, map data are obtained, grid generation processing is conducted on the map data, geographic grid features are extracted, the map data comprise one or more of point-shaped map data, linear map data and planar map data, and the geographic grid features comprise feature vectors describing various geographic indexes of each grid space; secondly, acquiring first telephone traffic data, and decomposing the first telephone traffic data according to the geographic grid characteristics to obtain second telephone traffic data; and finally, space traffic data and time sequence traffic data are generated according to the second traffic data and the geographic grid features, the space traffic data comprise data of traffic demand prediction of rasterized spatial distribution, and the time sequence traffic data comprise data of traffic demand prediction of normalized time sequence distribution. The method not only can support data-free cold start of energy-saving control of the wireless base station, but also can generate traffic demand prediction data of time-space domain decoupling according to geographic features and service modes of the wireless base station.

Description

Technical Field

[0001] The present application relates to the field of wireless communications, and in particular to a method and system for predicting traffic demand. Background Art

[0002] In the development of wireless communication technology, with the continuous increase and diversification of user needs, the management and optimization of wireless network resources are particularly important. Especially in the context of big data and artificial intelligence, wireless communication technology is increasingly concerned with how to improve service quality and reduce operating costs, while also meeting the needs of energy saving and environmental protection. Wireless traffic demand prediction is a key link in wireless communication technology. In the study of wireless traffic demand prediction, in order to achieve refined management of network resources, it is usually necessary to divide the geographic space into multiple grid units. These grid units can be a community, a large building, or a park. By accurately predicting the traffic demand of each grid unit, the energy use of the base station can be effectively optimized, thereby improving service quality and reducing operating costs. Summary of the invention

[0003] The present application provides a method and system for predicting traffic demand. The method and system for predicting traffic demand provided by the present application can not only support the cold start of energy-saving control of wireless base stations in the absence of traffic data, but also generate traffic demand predictions in the time domain and space domain according to the geographical characteristics of the base station and the correspondence between the traffic business model and the geographical indicators, further supporting the planning and layout of wireless base stations and energy-saving optimization control.

[0004] In the first aspect, a method for predicting traffic demand is proposed. First, map data is obtained, and the map data is subjected to raster generation processing and geographic raster features are extracted. The map data includes one or more of point map data, line map data, and surface map data. The geographic raster features include feature vectors that describe multiple geographic indicators of each raster space. Secondly, first traffic data is obtained, and the first traffic data is decomposed according to the geographic raster features to obtain second traffic data. Finally, spatial traffic data and time series traffic data are generated according to the second traffic data and the geographic raster features. The spatial traffic data includes data for predicting traffic demand for rasterized spatial distribution, and the time series traffic data includes data for predicting traffic demand for normalized time series distribution. Through the above three steps, the method provided in the present application generates a traffic flow forecast decoupled in the spatiotemporal domain and a time series forecast of the number of users according to the rasterized map data based on the building-level time series traffic flow decomposition and generation technology of spatiotemporal semantics.

[0005] First, the first step is introduced. In the above-mentioned method of performing raster generation processing on map data and extracting geographic raster features, the following steps may be included: first, a raster network is generated according to data anchor points, and the data anchor points include fine-grained traffic data as rasterization anchor points. Then, geographic raster features of the map data are extracted according to the raster network, and the geographic raster features include one or more of point raster features, line raster features, and surface raster features.

[0006] The specific method for extracting geographic grid features of rasterized map data mentioned above may include the following operations: when the map data includes point map data, extracting point grid features of the point map data, the point map data includes a plurality of different types of points of interest POIs distributed in the map space and a first geographic indicator of the point of interest, the first geographic indicator including altitude, longitude and latitude, and type of point of interest. Among them, the method for extracting point grid features of point map data includes counting the number of points of interest in each grid of the point map data, and combining one or more numbers of points of interest corresponding to each grid into point grid features.

[0007] The specific method for extracting geographic grid features of map data based on a grid network mentioned above may also include the following operations: when the map data includes linear map data, extracting linear grid features of the linear map data, the linear map data includes road network data distributed in the map space and a second geographic indicator of the road network data, the second geographic indicator including road length, road direction, and road geometry. Among them, the method for extracting linear grid features of linear map data includes calculating the shortest distance between the road inside each grid and the center point of the grid, and combining one or more shortest distances corresponding to each grid into a linear grid feature.

[0008] The specific method for extracting geographic grid features of map data based on a grid network mentioned above may also include the following operations: when the map data includes planar map data, extracting planar grid features of the planar map data, the planar map data includes a plurality of geometric figures describing building outlines distributed in the map space and a third geographic indicator of the geometric figures, the third geographic indicator including population, building area, land use type, and building purpose. Among them, the method for extracting planar grid features of planar map data includes counting the area proportion of the planar map data in each grid, and combining one or more area proportions corresponding to each grid into a planar grid feature.

[0009] In the case where the extracted geographic raster features include multiple types of point raster features, linear raster features and surface raster features, multiple types of the point raster features, linear raster features and surface raster features are combined to obtain a raster combination feature, the raster combination feature includes geographic indicators that describe multiple map data in each raster space, the geographic indicators include a first geographic indicator, a second geographic indicator or a third geographic indicator.

[0010] The method mentioned above for obtaining a grid combination feature by combining multiple types of point grid features, line grid features and surface grid features includes aligning and unifying multiple types of point grid features, line grid features and surface grid features to obtain a grid combination feature, the alignment and unification operation includes unification of coordinate reference systems, unification of grid coverage, unification of grid resolutions and unification of grid azimuths, and the geographic grid features include grid combination features.

[0011] Next, the second step is introduced. The method mentioned above for decomposing the first traffic data according to the geographic grid features to obtain the second traffic data includes the following steps. First, the first traffic data from different base stations or cells are decomposed according to the base station coverage model to obtain multiple cell traffic flows, and the cell traffic flows include the first traffic data in units of cells. Secondly, each cell traffic flow is decomposed into multiple groups of cell grid flows according to the geographic grid features according to the traffic decomposition model, and the cell grid flows include the fine-grained distribution pattern of the cell traffic flows that have been rasterized. Finally, multiple groups of cell grid flows are combined and converged to obtain the second traffic data.

[0012] The first traffic data includes time-series traffic flows from multiple base stations, and the second traffic data includes grid-distributed time-series traffic flows. The first traffic data also includes one or more of the industrial parameter data and user experience data; the second traffic data also includes one or more of the industrial parameter data and user experience data distributed in a grid.

[0013] Finally, the third step is introduced. The above-mentioned generation of spatial traffic data and time series traffic data based on the second traffic data and geographic grid features includes the following steps. First, the second traffic data and geographic grid features are input into the spatial generation model to generate spatial traffic data, wherein the spatial traffic data includes the rasterized distribution values ​​of the traffic demand forecast in the spatial domain. Multiple groups of clustered traffic flows are input into the time series generation model to generate multiple groups of time series traffic data, wherein the time series traffic data includes multiple groups of normalized distribution patterns of the traffic demand forecast in the time domain, and the multiple groups of clustered traffic flows include the traffic data classified according to the clustering discrimination results of the second traffic data.

[0014] The aforementioned inputting of multiple groups of clustered traffic flows into the time series generation model to generate multiple groups of time series traffic data may include the following steps. First, clustering analysis is performed on the second traffic data according to the geographic indicator type to obtain a clustering discrimination result, and the second traffic data is classified according to the clustering discrimination result to obtain multiple groups of clustered traffic flows, and the geographic indicator type includes multiple types of geographic indicators extracted from geographic grid features. Subsequently, the multiple groups of clustered traffic flows are respectively input into the time series generation model to generate multiple groups of time series traffic data, and the time series generation model includes an adversarial generative network model.

[0015] In some possible embodiments, the third step also includes grid-aligning the spatial traffic data and multiple groups of time-series traffic data and aggregating them to obtain spatiotemporal traffic data, which includes multi-mode traffic demand prediction data decoupled in the spatiotemporal domain.

[0016] In the second aspect, a traffic demand prediction device is proposed. It is characterized in that it includes a data processing and extraction module for obtaining map data, performing raster generation processing on the map data and extracting geographic raster features. The map data includes one or more of point map data, line map data and surface map data. The geographic raster features include feature vectors describing multiple geographic indicators of each raster space. It also includes a traffic space decomposition module for obtaining first traffic data and decomposing the first traffic data according to the geographic raster features to obtain second traffic data. It also includes a spatiotemporal data generation module for generating spatial traffic data and time series traffic data according to the second traffic data and the geographic raster features. The spatial traffic data includes data for predicting traffic demand for rasterized spatial distribution, and the time series traffic data includes data for predicting traffic demand for normalized time series distribution. The device provided in the present application generates a traffic flow forecast for spatiotemporal domain decoupled and a time series forecast for the number of users according to the rasterized map data through the above three modules based on the building-level time series traffic flow decomposition and generation technology of spatiotemporal semantics.

[0017] The first module is introduced first. The data processing and extraction module mentioned above is specifically used to generate a raster network based on data anchor points. The data anchor points include fine-grained traffic data as rasterization anchor points; and to extract geographic raster features of map data based on the raster network. The geographic raster features include one or more of point raster features, line raster features and surface raster features.

[0018] The data processing and extraction module mentioned above may include the following operations when extracting geographic grid features: when the map data includes point map data, extract the point grid features of the point map data, the point map data includes a plurality of different types of points of interest POIs distributed in the map space and a first geographic indicator of the point of interest, the first geographic indicator including altitude, longitude and latitude, and type of point of interest. Among them, the method for extracting the point grid features of the point map data includes counting the number of points of interest in each grid of the point map data, and combining one or more numbers of points of interest corresponding to each grid into a point grid feature.

[0019] The data processing and extraction module mentioned above may also include the following operations when extracting geographic grid features: when the map data includes linear map data, extracting linear grid features of the linear map data, the linear map data includes road network data distributed in the map space and a second geographic indicator of the road network data, the second geographic indicator includes road length, road direction and road geometry. Among them, the method for extracting linear grid features of linear map data includes calculating the shortest distance between the road inside each grid and the center point of the grid, and combining one or more shortest distances corresponding to each grid into a linear grid feature.

[0020] The data processing and extraction module mentioned above may also include the following operations when extracting geographic grid features: when the map data includes planar map data, extracting planar grid features of the planar map data, the planar map data includes multiple geometric figures describing building outlines distributed in the map space and third geographical indicators of the geometric figures, the third geographical indicators include population, building area, land use type and building purpose. Among them, the method for extracting planar grid features of planar map data includes counting the area proportion of the planar map data in each grid, and combining one or more area proportions corresponding to each grid into a planar grid feature.

[0021] When the extracted geographic raster features include multiple types of point raster features, linear raster features and surface raster features, the data processing and extraction module combines multiple types of point raster features, linear raster features and surface raster features to obtain raster combination features, where the raster combination features include geographic indicators that describe multiple map data in each raster space, and the geographic indicators include first geographic indicators, second geographic indicators or third geographic indicators.

[0022] The method for the data processing and extraction module mentioned above to obtain a grid combination feature by combining multiple types of point grid features, linear grid features and surface grid features includes aligning and unifying multiple types of point grid features, linear grid features and surface grid features to obtain a grid combination feature, the alignment and unification operation includes unification of coordinate reference system, unification of grid coverage, unification of grid resolution and unification of grid azimuth, and the geographic grid features include grid combination features.

[0023] Next, the second module is introduced. In the above-mentioned traffic space decomposition module, it is first used to decompose the first traffic data from different base stations or cells according to the base station coverage model to obtain multiple cell traffic flows. The cell traffic flows include the first traffic data in units of cells. Secondly, it is used to decompose each cell traffic flow into multiple groups of cell grid flows according to the geographic grid features according to the traffic decomposition model. The cell grid flows include the fine-grained distribution pattern of the cell traffic flows after rasterization. Finally, it is used to combine and converge multiple groups of cell grid flows to obtain the second traffic data. Among them, the first traffic data includes time-series traffic flows from multiple base stations, and the second traffic data includes time-series traffic flows distributed in a grid.

[0024] Finally, the third module is introduced. In the above-mentioned spatiotemporal data generation module, it is specifically used to input the second traffic data and geographic grid features into the spatial generation model to generate spatial traffic data, wherein the spatial traffic data includes the grid distribution value of the traffic demand forecast in the spatial domain. And it is used to input multiple groups of clustered traffic flows into the time series generation model to generate multiple groups of time series traffic data, wherein the time series traffic data includes multiple groups of normalized distribution patterns of traffic demand forecast in the time domain, and the multiple groups of clustered traffic flows include the flow data classified according to the cluster discrimination results of the second traffic data.

[0025] The aforementioned spatiotemporal data generation module generates multiple groups of time series traffic data, specifically including the following operations: first, clustering the second traffic data according to the geographic indicator type to obtain a clustering discrimination result, and classifying the second traffic data according to the clustering discrimination result to obtain multiple groups of clustered traffic flows, and the geographic indicator type includes multiple types of geographic indicators extracted from geographic grid features. Subsequently, the multiple groups of clustered traffic flows are respectively input into the time series generation model to generate multiple groups of time series traffic data, and the time series generation model includes an adversarial generation network model.

[0026] In some possible embodiments, the spatiotemporal data generation module is also used to grid-align the spatial traffic data and multiple groups of time-series traffic data and aggregate them to obtain spatiotemporal traffic data, which includes multi-mode traffic demand forecasting data decoupled from the spatiotemporal domain.

[0027] In a third aspect, a data processing device is provided, comprising a memory and a processor, wherein the memory stores instructions, and when the processor executes the instructions in the memory, the method of any embodiment of the first aspect can be implemented.

[0028] In a fourth aspect, a non-volatile storage medium is proposed, comprising a plurality of instructions, which can implement the method required by any embodiment of the first aspect when the instructions are executed by a data processing device. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0030] Figure 1 It is a brief schematic diagram of a signal coverage model of a wireless communication base station provided by the present application;

[0031] Figure 2 It is a schematic diagram of a traffic demand prediction system provided by the present application;

[0032] Figure 3 It is a schematic diagram of a data processing and extraction module provided by the present application;

[0033] Figure 4 A schematic diagram of a traffic space decomposition module provided in this application;

[0034] Figure 5 A schematic diagram of a spatiotemporal data generation module provided in this application;

[0035] Figure 6 A schematic diagram of another traffic demand forecasting system provided for this application;

[0036] Figure 7 A flow chart of a method for predicting traffic demand provided by this application;

[0037] Figure 8 A flowchart of a data processing and extraction method provided in this application;

[0038] Fig. 9 A flow chart of a traffic space decomposition method provided in this application;

[0039] Fig.10 A schematic diagram of a process for generating spatiotemporal data provided in this application;

[0040] Fig.11 A schematic diagram of a traffic demand prediction device provided by the present application;

[0041] Fig.12A structural schematic diagram of a data processing device provided in this application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described through the implementation method with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0043] In the description of this application, unless otherwise clearly defined, words such as "setting" should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meaning of the above words in this application based on the specific content of the technical solution.

[0044] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific embodiments", or "some examples" means that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0045] In recent years, with the continuous development of the deployment of wireless communication base stations and the optimization of wireless communication networks, new requirements have been put forward for the enhancement of communication performance, expansion of coverage and reduction of energy consumption of wireless communication base stations. Wireless traffic demand prediction is a key link in wireless communication technology. The traffic flow data generated by wireless traffic demand prediction technology can assist in guiding base station deployment planning, which can not only reduce construction and operation costs and improve network efficiency, but also help energy conservation and carbon reduction of information and communication infrastructure, and achieve energy conservation and emission reduction goals in the communication industry. Therefore, wireless traffic demand prediction technology has attracted attention in recent years as a hot wireless communication technology with application value. In order to facilitate the understanding of the present invention by people in related fields, the application scenarios to which this application is applicable are briefly described in combination with embodiments and drawings.

[0046] See also Figure 1 , Figure 1 A schematic diagram of a signal coverage model of a wireless communication base station provided in this application. Figure 1 The signal coverage model of the wireless communication base station shown includes a wireless base station facility 110 , a base station signal range 120 , point map data 130 , line map data 140 , and surface map data 150 .

[0047] First, the wireless base station facility 110 is introduced. The signal coverage model of this embodiment includes two wireless base station facilities 110. Figure 1 It can be seen intuitively that the wireless base station facility 110 is a macro base station with a large coverage area. Figure 1 The wireless base station facility 110 shown is only used as an example. In other possible embodiments, the wireless base station facility 110 introduced by the method of the present application can also be a wireless base station communication facility system composed of one or more of a macro base station, a micro base station (microsite), a picosite, a femtosite and a cellular base station (cellular site). Since the planar map data 150 included in this embodiment is used to represent different types of building units, in some possible embodiments, the wireless base station facility 110 not only includes the above-mentioned wireless base station equipment commonly found outdoors, but also includes a base station distribution system located around the building represented by the planar map data 150. Among them, the base station distribution system can include an indoor distribution system for indoor wireless signal coverage, an outdoor distribution system for deep coverage of cell wireless signals, and an indoor and outdoor distribution system for indoor and outdoor wireless signal coverage of buildings. In addition, the type of the wireless base station facility 110 can also be other wireless communication facilities including massive MIMO (Massive Multiple Input Multiple Output) antennas and a combination of multiple of the above facilities. The number of wireless base station facilities 110 can also be less or more, which is not limited here.

[0048] Next, the base station signal range 120 is introduced. In this embodiment, a simple circle with two wireless base station facilities 110 as the center represents the wireless signal coverage range of the wireless base station facilities 110. In other possible embodiments, depending on the number and type complexity of the wireless base station facilities 110, the base station signal range 120 can also have signal coverage characteristics of different shapes according to actual situations. Figure 1 It can be seen that the point map data 130 , the line map data 140 , and the surface map data 150 included in this embodiment are all included in the base station signal range 120 .

[0049] The point map data 130 is introduced again. The point map data 130 includes a plurality of different types of points of interest (POI) distributed in the map space and a first geographical indicator corresponding to the description of the POI, including altitude, longitude and latitude, type of point of interest, etc. In some possible embodiments, the point map data 130 may be a data set of one or more map coordinate points that are highlighted because of the interest of the data annotator.

[0050] Next, the linear map data 140 is introduced, which includes road network data distributed in the map space and second geographical indicators corresponding to the road network data, including road length, direction, geometric shape, etc. In some possible embodiments, the linear map data 140 may be a data set representing national roads, provincial roads, county roads, expressways, municipal roads, railways, bridges, tunnels, culverts, etc., for describing the map road network.

[0051] Finally, the planar map data 150 is introduced. The planar map data includes a plurality of geometric figures describing building outlines distributed in the map space and third geographical indicators of the geometric figures including population, building area, land use type, and building purpose, etc. In some possible embodiments, the planar map data 150 may be a data set consisting of one or more areas of interest (AOI) representing residential areas, commercial clusters, large venues, transportation hubs, industrial parks, park scenic areas, administrative office buildings, hospitals, schools, factories, village buildings, etc., which have large traffic flows and high wireless communication requirements.

[0052] In the signal coverage model introduced in this embodiment, the traffic sources represented by the point map data 130, the line map data 140 and the surface map data 150 send the traffic demand to the wireless base station facility 110. The wireless base station facility 110 also collects wireless communication signals from various traffic sources within the base station signal range 120. Therefore, the traffic demand prediction method with the base station as the core has the advantage of low traffic statistics cost.

[0053] Existing traffic demand prediction methods rely on historical data, and their prediction results are often unsatisfactory when faced with missing data or large changes in data. Next, two existing technologies related to the present invention and their shortcomings are introduced.

[0054] First, we will introduce the method of predicting traffic flow based on time series. There are usually two solutions in the existing technology. The first idea is a solution based on time series prediction, which predicts future time series traffic flow data through historical time series traffic flow data. The advantage of this solution is that it has high accuracy in future predictions and the technical solution is relatively mature. The problem with this solution is that it requires long-term historical training data to achieve accurate prediction results. It is not very useful when there is no long-term historical data available in the scenario. The second idea is a solution based on time series generation, which generates data on future traffic flow by establishing a model. The advantage of this solution is that it does not require long-term historical data and is applicable to a wide range of scenarios. The problem with this solution is that most studies focus on medical or traffic time series, lack analysis methods for wireless base station data, and require a large amount of training data to support the training of the generation model.

[0055] It can be seen that for the first technology - the method of predicting traffic based on time series, the existing base station traffic prediction models based on time series prediction require long-term training data to achieve practical accuracy, and are difficult to apply to base station planning, deployment and other scenarios. At the same time, most of the existing generation technologies are aimed at traffic scenarios, and require large-scale training data to support the generation model. There is no effective solution that can be applied to the spatiotemporal generation problem of telecommunications base station data.

[0056] Then, the method of decomposing the traffic flow based on time series is introduced. There are usually two solutions in the existing technology. The first idea is a solution based on trend-season decomposition. The advantage of this solution is that the time series can be decomposed into explainable seasonal, trend and residual components. The problem with this solution is that it only considers the decomposition at the time level but not the factors of geographic space, and does not have the decomposition function of geographic entities. The second idea is a solution based on spatial regional decomposition. The advantage of this solution is that it can analyze the spatial geographic entities within the coverage area of ​​the base station and explore the inherent traffic patterns of areas with similar geographical functions. The problem with this solution is that most algorithms for spatial decomposition do not consider the correlation at the time level, and lack analysis methods for wireless base station data.

[0057] It can be seen that for the second technology - time series decomposition of traffic flow method, most of the existing time series decomposition algorithms are based on trend and seasonal components, which are not suitable for the spatial decomposition task in this problem. At the same time, most of the existing spatial decomposition algorithms only consider algorithms based on time series patterns, and there is no mature solution for urban base stations.

[0058] In order to solve the above problems, the present invention proposes a traffic demand prediction method and system. The method of the present application is based on the building-level time-series traffic flow decomposition and generation technology of spatiotemporal semantics, and generates a traffic flow forecast and a user number time series forecast that are decoupled in the spatiotemporal domain according to rasterized map data. The traffic demand prediction method and system provided in the present application can not only support the cold start of energy-saving control of wireless base stations in the absence of traffic data, but also generate traffic demand forecasts in the time domain and space domain according to the geographical characteristics of the base station and the correspondence between the traffic business model and the geographical indicators, further supporting the planning and layout of wireless base stations and energy-saving optimization control.

[0059] See also Figure 2 , Figure 2 The diagram of a traffic demand prediction system provided by the present application is as follows: The traffic demand prediction system provided by the present application includes three modules, namely a data processing and extraction module 210 , a traffic space decomposition module 220 and a spatiotemporal data generation module 230 .

[0060] First, the first module is introduced, regarding the functions of the data processing and extraction module 210, including obtaining map data 201, rasterizing the map data 201, and extracting geographic grid features 207 of the map data 201. The map data 201 includes one or more of point map data, line map data, and surface map data. The geographic grid features 207 include feature vectors that describe multiple geographic indicators of each grid space. The map data 201 of this embodiment may correspond to Figure 1 The point map data 130, the line map data 140 and the area map data 150 in the illustrated embodiment. Specifically, the data processing and extraction module 210 first performs rasterization processing on the map data 201 according to the raster distribution characteristics, and the raster distribution characteristics include raster data obtained by analyzing the pre-rasterized data anchor points 202. The data processing and extraction module 210 then extracts the geographic raster features 207 of the rasterized map data 201, wherein the geographic raster features 207 include one or more of the point raster features, the line raster features and the area raster features.

[0061] In order to help people in the relevant field better understand the functions of the data processing and extraction module 210, the detailed operation of the data processing and extraction module 210 is introduced below with a specific embodiment.

[0062] See also Figure 3 , Figure 3 A schematic diagram of a data processing and extraction module provided in this application. Figure 3 The functions of the data processing and extraction module shown mainly include grid generation processing 310 , grid feature extraction 320 , and alignment and unification operation 330 .

[0063] First, the grid generation process 310 is introduced. The data processing and extraction module generates a grid network 302 according to the data anchor point 301. In some possible situations, the data anchor point 301 can be the grid data obtained by generating fine-grained traffic data used as a rasterization anchor point. These fine-grained traffic data can be composed of fine-grained traffic flows that have been rasterized by the traffic space decomposition module 220. In some possible situations, the function of generating the data anchor point 301 without data can also be realized by pre-setting the grid distribution characteristics.

[0064] Next, the grid feature extraction 320 is introduced. The data processing and extraction module extracts the geographic grid features 308 of the map data 303 according to the grid network 302, including one or more of the point grid features 304, the line grid features 305 and the surface grid features 306. For ease of understanding, Figure 3A situation in which all three types of raster features exist is given. In other possible embodiments, depending on the type of data contained in the map data 303, the geographic raster feature 308 may include a form composed of one or two of the above-mentioned raster features, and may also include the raster combination feature 307 described below, which is not limited here.

[0065] In the case where the map data 303 includes point map data, point grid features 304 of the point map data are extracted, and the point map data includes a plurality of different types of points of interest POIs distributed in the map space and a first geographical indicator of the points of interest, and the first geographical indicator includes altitude, longitude and latitude, and type of the point of interest. The calculation method of the point grid features 304 is specifically to count the number of points of interest in each grid of the point map data, and combine one or more numbers of points of interest corresponding to each grid into point grid features.

[0066] In the case where the map data 303 includes linear map data, linear grid features 305 of the linear map data are extracted, and the linear map data includes road network data distributed in the map space and a second geographical indicator of the road network data, and the second geographical indicator includes the length, direction and geometric shape of the road. The calculation method of the linear grid feature 305 is specifically to calculate the shortest distance between the road inside each grid and the center point of the grid, and combine one or more shortest distances corresponding to each grid into a linear grid feature.

[0067] In the case where the map data 303 includes planar map data, planar grid features 306 of the planar map data are extracted, and the planar map data includes a plurality of geometric figures describing building outlines distributed in the map space and a third geographical indicator of the geometric figures, and the third geographical indicator includes population, building area, land use type, and building purpose. The calculation method of the planar grid features 306 is specifically to count the area proportion of the planar map data in each grid, and combine one or more area proportions corresponding to each grid into a planar grid feature.

[0068] Finally, the alignment and unification operation 330 is introduced, including coordinate reference system unification 331, grid coverage unification 332, grid resolution unification 333, and grid azimuth unification 334. In the case where the extracted geographic grid features 308 include multiple types of point grid features 304, linear grid features 305, and planar grid features 306, the data processing and extraction module performs an alignment and unification operation 330 on multiple types of point grid features 304, linear grid features 305, and planar grid features 306 to obtain grid combination features 307, and the grid combination features 307 include geographic indicators that describe multiple map data in each grid space, and the geographic indicators include the first geographic indicator, the second geographic indicator, or the third geographic indicator mentioned above.

[0069] Through the above three steps, the data feature extraction module can output geographic grid features 308 in one or more forms including point grid features 304, line grid features 305, surface grid features 306, and grid combination features 307. It can be seen that Figure 3 The geographic grid feature 308 shown corresponds to Figure 2 Geographic raster feature 207 is shown.

[0070] Next, the second module is introduced, regarding the function of the traffic space decomposition module 220, including obtaining the first traffic data 206, and decomposing the first traffic data 206 according to the geographic grid feature 207 to obtain the second traffic data 208. Among them, the first traffic data 206 includes the time-series traffic flow 203 from multiple wireless base stations, and the second traffic data 208 includes the grid-distributed time-series traffic flow 203. In other possible embodiments, the first traffic data 206 also includes one or more of the engineering parameter data 205 and the user experience data 204, and the second traffic data also includes one or more of the grid-distributed engineering parameter data 205 and the user experience data 204. The engineering parameter data 205 is about the engineering parameters of the wireless base station antenna, mainly including antenna height, antenna pitch angle, antenna azimuth angle, antenna position, antenna radiation parameters, antenna circuit parameters, etc. These parameters have a decisive influence on the electromagnetic coverage of the base station. The user experience data 204 includes traffic type data set based on the user of the traffic flow as the basic unit.

[0071] In order to help people in the relevant field better understand the function of the traffic space decomposition module 220, the detailed operation of the traffic space decomposition module 220 is introduced below with a specific embodiment.

[0072] See also Figure 4 , Figure 4 A schematic diagram of a traffic space decomposition module provided in this application. Figure 4 The traffic spatial decomposition module shown mainly comprises a base station coverage model 410 , a traffic decomposition model 420 and a grid aggregation operation 430 .

[0073] First, the traffic space decomposition module decomposes the first traffic data 401 from different wireless base stations, cells or buildings according to the base station coverage model 410 to obtain multiple cell traffic flows 402, and the cell traffic flows 402 include the first traffic data 206 in cells. For more information about the base station coverage model 410, see Figure 6Specific contents of the embodiment shown. Secondly, the traffic space decomposition module decomposes each cell traffic flow 402 into multiple groups of cell grid traffic 404 according to the geographic grid feature 403 according to the traffic decomposition model 420, and the cell grid traffic 404 includes a fine-grained distribution pattern of the cell traffic flow 402 after rasterization. Finally, the traffic space decomposition module combines and aggregates the multiple groups of cell grid traffic 404 through the grid aggregation operation 430 to obtain the second traffic data 405.

[0074] Finally, the third module is introduced, which is about the function of the spatiotemporal data generation module 230, including generating spatial traffic data 231 and time series traffic data 232 according to the second traffic data 208 and the geographic grid feature 207. Among them, the spatial traffic data 231 includes the data of the traffic demand prediction of the gridded spatial distribution, and the time series traffic data 232 includes the data of the traffic demand prediction of the normalized time series distribution.

[0075] In order to help people in related fields better understand the functions of the spatiotemporal data generation module 230, the detailed operation of the spatiotemporal data generation module 230 is introduced below with a specific embodiment.

[0076] See also Figure 5 , Figure 5 A functional schematic diagram of a spatiotemporal data generation module provided in this application. Figure 5 The composition of the spatiotemporal data generation module shown mainly includes a cluster analysis algorithm 510 , a time series generation model 520 and a space generation model 530 .

[0077] First, the method for obtaining the time series traffic data 506 is introduced. In this embodiment, the spatiotemporal data generation module obtains the time series traffic data 506 through the second traffic data 501 and the geographic grid feature 502 obtained by the two modules mentioned above. In the first step, the spatiotemporal data generation module obtains the cluster discrimination result of the second traffic data 501 through the cluster analysis algorithm 510 according to the geographic indicator type 503, and classifies the second traffic data according to the cluster discrimination result to obtain multiple groups of cluster traffic flow 504. The geographic indicator type 503 includes a type set of multiple geographic indicators extracted from the geographic grid feature 502. The types of these geographic indicators may include indicators such as altitude, longitude and latitude, point of interest type, road length, road direction, road geometry, population, building area, land use type and building use, etc., which are used to describe the geographic characteristics of map data. In the second step, the spatiotemporal data generation module inputs the multiple groups of cluster traffic flow 504 into the time series generation model 520 to generate multiple groups of time series traffic data 506. When the time series generation model is an adversarial generation network model, the input of this step may also include normalized distribution noise 505.

[0078] Next, a method for acquiring the spatial traffic data 507 is introduced. In this embodiment, the spatiotemporal data generation module acquires the spatial traffic data 507 through the second traffic data 501 and the geographic grid features 502 obtained by the two modules described above.

[0079] It can be seen that this embodiment obtains the time-series traffic data 506 and the space-series traffic data 507 in no particular order. At the same time, this embodiment not only includes obtaining the time-series traffic data 506 and the space-series traffic data 507, but also includes grid-aligning the obtained space-series traffic data 507 and multiple groups of time-series traffic data 506 and aggregating them to obtain spatiotemporal traffic data 508. The spatiotemporal traffic data includes data on multi-mode traffic demand prediction with decoupling of the spatiotemporal domain.

[0080] In order to help people in related fields better understand the functions of the traffic demand prediction system, the operation of the traffic demand prediction system is introduced in a specific embodiment through the base station coverage theory.

[0081] See also Figure 6 , Figure 6 A schematic diagram of another traffic demand forecasting system provided in this application. Figure 6 The traffic demand prediction system shown in the figure mainly includes a data processing and extraction module 610, a traffic space decomposition module 620, and a spatiotemporal data generation module 630. This embodiment is to highlight the graphical data form of the wireless base station coverage theoretical model. As an example, only map data 601, grid geographic features 602, second traffic data 603, and spatiotemporal traffic data 604 are listed. For other data that realize the functions of the traffic demand prediction system, please refer to Figures 2 to 6 The detailed introduction is not repeated here.

[0082] For details about the map data 601, see Figure 1The detailed description of the embodiment shown is omitted. The map data 601 obtains the rasterized distributed raster geographic features 602 through the relevant operations of the data processing and extraction module 610. For each grid of the raster geographic features 602, one or more raster features of geographical indicators may be included. The raster geographic features 602 obtain the second traffic data 603 through the relevant operations of the traffic space decomposition module 620. For each grid of the second traffic data 603, it includes a fine-grained traffic flow distribution pattern obtained by aggregating after decomposing the first traffic data composed of multiple known traffic flow anchor points. The second traffic data 603 obtains the spatiotemporal traffic data 604 through the relevant operations of the spatiotemporal data generation module 630. The spatiotemporal traffic data 604 is obtained by the time series traffic data and spatial traffic data generated by the spatiotemporal data generation module 630, including multiple groups of traffic flow prediction data decoupled in time and space. For each group of traffic flow prediction data, each grid includes a traffic flow prediction generated in the entire spatial domain. In each group of traffic flow prediction data, since the prediction value of each grid is static, there is no concept of time domain. Through the multiple groups of traffic flow prediction data shown in the spatiotemporal traffic data 604, a traffic flow prediction model in the time domain can be constructed. This flow prediction model is consistent with the time series traffic data and is normalized in the time domain.

[0083] Through Figures 2 to 6 The embodiment shown introduces a traffic demand prediction system provided by the present application. Next, a traffic demand prediction method provided by the present application is introduced through specific operation steps.

[0084] See also Figure 7 , Figure 7 The flow chart of a method for predicting traffic demand provided by the present application is shown in FIG. The method for predicting traffic demand shown in this embodiment can be used to implement the following: Figure 2 The traffic demand prediction system of the embodiment shown has the following specific steps:

[0085] S101: Acquire map data, perform raster generation processing on the map data and extract geographic raster features.

[0086] This step introduces the data processing and extraction method, including obtaining map data, rasterizing the map data, and extracting geographic raster features of the map data. The map data includes one or more of point map data, line map data, and surface map data. The geographic raster features include feature vectors that describe multiple geographic indicators of each raster space. The map data of this embodiment may correspond to Figure 1Point map data, line map data and surface map data in the illustrated embodiment. Specifically, the data processing and extraction module first rasterizes the map data according to the raster distribution characteristics, and the raster distribution characteristics include raster data obtained by analyzing the pre-rasterized data anchor points. The data processing and extraction module then extracts the geographic raster features of the rasterized map data, wherein the geographic raster features include one or more of point raster features, line raster features and surface raster features.

[0087] For ease of understanding, the following is an example to introduce the detailed operation of the data processing and extraction method. Figure 8 , Figure 8 This is a flow chart of a data processing and extraction method provided in this application. The specific steps are as follows.

[0088] S201: Generate a grid network according to data anchor points.

[0089] This step generates a grid network based on the data anchor point. In some possible situations, the data anchor point can be the grid data obtained by generating fine-grained traffic data used as the rasterized anchor point. These fine-grained traffic data can be composed of fine-grained traffic flows that have been rasterized by the traffic space decomposition module. In some possible situations, the function of generating data anchor points without data can also be realized through pre-set grid distribution features.

[0090] S202: Extracting geographic grid features of map data according to the grid network.

[0091] This step extracts the geographic grid features of the map data based on the grid network, including one or more of point grid features, line grid features, and surface grid features. For ease of understanding, Figure 3 A situation in which all three types of raster features exist is given. In other possible embodiments, depending on the type of data contained in the map data, the geographic raster feature may include a form composed of one or two of the above-mentioned raster features, and may also include the raster combination feature described below, which is not limited here.

[0092] S203: When the extracted geographic grid features include multiple types of point grid features, linear grid features, and surface grid features, multiple types of the point grid features, linear grid features, and surface grid features are combined to obtain grid combination features.

[0093] In this step, multiple types of point grid features, line grid features, and surface grid features are aligned and unified to obtain grid combination features. The alignment and unification operations include coordinate reference system unification, grid coverage unification, grid resolution unification, and grid azimuth unification. The grid combination features include geographic indicators that describe multiple map data in each grid space. The geographic indicators include the first geographic indicator, the second geographic indicator, or the third geographic indicator mentioned above.

[0094] To save space, for more information on data processing and extraction methods, please refer to Figure 2 , Figure 3 as well as Figure 6 The detailed introduction of the relevant functional operations of the data extraction module in the illustrated embodiment will not be repeated here.

[0095] S102: Acquire first traffic data, and decompose the first traffic data according to geographic grid features to obtain second traffic data.

[0096] This step introduces a method for spatial decomposition of traffic, including obtaining first traffic data, and decomposing the first traffic data according to geographic grid features to obtain second traffic data. The first traffic data includes time-series traffic flows from multiple wireless base stations, and the second traffic data includes grid-distributed time-series traffic flows. In other possible embodiments, the first traffic data also includes one or more of industrial parameter data and user experience data, and the second traffic data also includes one or more of grid-distributed industrial parameter data and user experience data.

[0097] For ease of understanding, the following is an example to introduce the detailed steps of the traffic space decomposition method. Fig. 9 , Fig. 9 This is a flow chart of a traffic space decomposition method provided by this application. The specific steps are as follows.

[0098] S301: Decomposing first traffic data from different base stations or cells according to a base station coverage model to obtain multiple cell traffic flows.

[0099] In this step, the first traffic data from different wireless base stations, cells or buildings are decomposed into multiple cell traffic flows according to the base station coverage model. The cell traffic flows include the first traffic data in cells. For more information about the base station coverage model, see Figure 6 The specific contents of the embodiment shown.

[0100] S302: Decomposing the traffic flow of each cell into multiple groups of cell grid flows according to the geographic grid features based on the traffic decomposition model.

[0101] This step decomposes each cell traffic flow into multiple groups of cell grid flows according to the geographic grid features based on the traffic decomposition model. The cell grid flows include fine-grained distribution patterns of cell traffic flows that have been gridded.

[0102] S303: Aggregate multiple groups of cell grid traffic to obtain second traffic data.

[0103] This step aggregates multiple groups of cell grid traffic combinations through a grid aggregation operation to obtain second traffic data.

[0104] To save space, more information about the method of spatial decomposition of traffic can be found in Figure 2 , Figure 4 as well as Figure 6 The detailed introduction of the functional operations related to the traffic space decomposition module in the illustrated embodiment will not be repeated here.

[0105] S103: Generate spatial traffic data and temporal traffic data according to the second traffic data and geographic grid features.

[0106] This step introduces a method for generating spatiotemporal data, including generating spatial traffic data and time series traffic data according to the second traffic data and geographic grid features. The spatial traffic data includes data for predicting traffic demand of gridded spatial distribution, and the time series traffic data includes data for predicting traffic demand of normalized time series distribution.

[0107] For ease of understanding, the following is an example to introduce the detailed steps of the spatiotemporal data generation method. Fig.10 , Fig.10 This is a flow chart of a method for generating spatiotemporal data provided by this application. The specific steps are as follows.

[0108] S401: Inputting the second traffic data and geographic grid features into a spatial generation model to generate spatial traffic data.

[0109] In this embodiment, this step acquires spatial traffic data through the second traffic data obtained by the two modules above and the geographic grid features.

[0110] In this step, the timing traffic data and the spatial traffic data are obtained in no particular order. Therefore, in other possible embodiments, the order of S401 and S402 can be interchanged, which is not limited here.

[0111] S402: Input multiple groups of clustered traffic flows into a time series generation model to generate multiple groups of time series traffic data.

[0112] This step obtains time series traffic data through the second traffic data obtained by the two modules in the previous text and the geographic grid features. In the first step, the spatiotemporal data generation module obtains cluster discrimination results by using the cluster analysis algorithm for the second traffic data according to the geographic indicator type, and classifies the second traffic data according to the cluster discrimination results to obtain multiple groups of cluster traffic flows. The geographic indicator type includes a set of types of multiple geographic indicators extracted from the geographic grid features. These types of geographic indicators may include indicators such as altitude, longitude and latitude, point of interest type, road length, road direction, road geometry, population, building area, land type, and building use, etc., which are used to describe the geographic characteristics of map data. In the second step, the spatiotemporal data generation module inputs multiple groups of cluster traffic flows into the time series generation model to generate multiple groups of time series traffic data. In the case where the time series generation model is an adversarial generation network model, the input of this step may also include normalized distributed noise.

[0113] The spatiotemporal data generation module in this embodiment obtains the time-series traffic data and the spatial traffic data in no particular order. Therefore, in other possible embodiments, the order of S401 and S402 can be interchanged, which is not limited here.

[0114] S403: Grid-align the spatial traffic data and multiple groups of temporal traffic data and aggregate them to obtain spatiotemporal traffic data.

[0115] In this embodiment, this step also includes grid-aligning the obtained spatial traffic data and multiple groups of time-series traffic data and aggregating them to obtain spatiotemporal traffic data, which includes multi-mode traffic demand prediction data decoupled in the spatiotemporal domain.

[0116] To save space, more information about the methods of generating spatiotemporal data can be found in Figure 2 , Figure 5 as well as Figure 6 The detailed introduction of the functional operations related to the traffic space decomposition module in the illustrated embodiment will not be repeated here.

[0117] The traffic demand prediction method shown in this application generates a traffic flow prediction and a user number time series prediction that are decoupled in the spatiotemporal domain based on rasterized map data through a building-level time-series traffic flow decomposition and generation technology based on spatiotemporal semantics. The traffic demand prediction method and system provided in this application can not only support the cold start of energy-saving control of wireless base stations in the absence of traffic data, but also generate traffic demand predictions in the time domain and space domain based on the geographical characteristics of the base station and the correspondence between the traffic business model and the geographical indicators, further supporting the planning and layout of wireless base stations and energy-saving optimization control.

[0118] See also Fig.11 , Fig.11The diagram of a traffic demand prediction device provided by the present application includes a data processing and extraction module 1110 , a traffic space decomposition module 1120 and a spatiotemporal data generation module 1130 .

[0119] The data processing and extraction module 1110 is used to obtain map data, rasterize the map data, and extract geographic raster features of the map data. The map data includes one or more of point map data, line map data, and surface map data. The geographic raster features include feature vectors that describe multiple geographic indicators of each raster space. The map data of this embodiment may correspond to Figure 1 Point map data, line map data and surface map data in the illustrated embodiment. Specifically, the data processing and extraction module first rasterizes the map data according to the raster distribution characteristics, and the raster distribution characteristics include raster data obtained by analyzing the pre-rasterized data anchor points 202. The data processing and extraction module then extracts the geographic raster features of the rasterized map data 201, wherein the geographic raster features include one or more of point raster features, line raster features and surface raster features.

[0120] The traffic space decomposition module 1120 is used to obtain the first traffic data, and decompose the first traffic data according to the geographic grid features to obtain the second traffic data. The first traffic data includes time-series traffic flows from multiple wireless base stations, and the second traffic data includes grid-distributed time-series traffic flows. In other possible embodiments, the first traffic data also includes one or more of the engineering parameter data and the user experience data, and the second traffic data also includes one or more of the grid-distributed engineering parameter data and the user experience data. The engineering parameter data is about the engineering parameters of the wireless base station antenna, mainly including antenna height, antenna pitch angle, antenna azimuth angle, antenna position, antenna radiation parameters, antenna circuit parameters, etc. These parameters have a decisive influence on the electromagnetic coverage of the base station. The user experience data includes traffic type data set in the cell range with users of traffic flow as the basic unit.

[0121] The spatiotemporal data generation module 1130 is used to generate spatial traffic data and time series traffic data according to the second traffic data and geographic grid features. The spatial traffic data includes data of traffic demand prediction of gridded spatial distribution, and the time series traffic data includes data of traffic demand prediction of normalized time series distribution. It can be understood that the spatiotemporal data generation module 1130 obtains the time series traffic data and the spatial traffic data in no particular order. At the same time, this embodiment not only includes obtaining the time series traffic data and the spatial traffic data, but also includes grid alignment of the obtained spatial traffic data and multiple groups of time series traffic data and aggregating them to obtain the spatiotemporal traffic data. The spatiotemporal traffic data includes data of multi-mode traffic demand prediction with decoupling of the spatiotemporal domain.

[0122] For more relevant concepts and detailed descriptions of traffic demand prediction introduced in this embodiment, please refer to Figures 2 to 10 The descriptions of the traffic demand prediction system and the traffic demand prediction method shown are not repeated here to save space.

[0123] See also Fig.12 , Fig.12 1 is a schematic diagram of the structure of a data processing device provided by the present application. The data processing device shown in the present application includes: one or more first processors 1210, a data collector 1220, a communication interface 1230, an external memory 1240 and a bus 1250. Among them, the first processor 1210, the data collector 1220, the communication interface 1230, and the external memory 1240 can be connected through the bus 1250.

[0124] The data collector 1220 is used to collect relevant data generated during the construction of the traffic demand prediction method as described in steps S101 to S103, steps S201 to S203, steps S301 to S303 and steps S401 to S403, as well as data acquisition between various modules.

[0125] The first processor 1210 includes one or more general-purpose processors, wherein the general-purpose processor can be any type of device capable of processing electronic instructions, including a central processing unit (CPU), a microprocessor, a microcontroller, a main processor, a controller, and an ASIC (Application Specific Integrated Circuit), etc. The first processor 1210 executes various types of digital storage instructions, such as software or firmware programs stored in the external memory 1240, which enables the data processing device to provide a wide variety of services. For example, the first processor 1210 can execute programs or process data to perform at least a portion of the methods discussed herein.

[0126] The communication interface 1230 may be a wired interface (eg, an Ethernet interface) for communicating with other data processing devices or users. The related methods and concepts for implementation may refer to the detailed introduction as described above and will not be repeated here.

[0127] The external memory 1240 may include a volatile memory, such as a random access memory (RAM); the memory may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD). The memory may also include a combination of the above-mentioned types of memory. The external memory 1240 may store program code and program data.

[0128] The first processor 1210 can be used to execute the method contents described in steps S101 to S103, steps S301 to S303, and steps S401 to S403 by calling the program code in the external memory 1240. For the sake of simplicity, the detailed description of this embodiment can be specifically referred to Figures 2 to 10 Relevant description and specific text content of the traffic demand prediction system and traffic demand prediction method.

[0129] Although the present invention describes functions performed by certain virtual components and physical entities, in a cloud service system application scenario, some or all of the above-mentioned encoding relationship management process may be distributed among multiple virtual components and entities, and one or more instances of the process may be executed on a distributed system, a virtual machine, and one or more cloud service systems.

[0130] The present application provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, all the steps of the above-mentioned traffic demand prediction method and system are implemented. For example, when the processor executes the computer program, the method process and related operations such as steps S101 to S103, steps S201 to S203, steps S301 to S303 and steps S401 to S403 are implemented.

[0131] The computer software product is stored in a storage medium, including several instructions for making a computer device perform all or part of the steps of the method described in each embodiment of the present application. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server, cloud service center or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server, cloud service center or data center.

[0132] The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media, or a cloud storage resource provided by a cloud server. The available medium may be a magnetic medium (e.g., a floppy disk, a storage disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk, SSD), etc.

[0133] In addition, Fig.12 The logic instructions in the external memory 1240 shown can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product.

[0134] The present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements all the steps of the above-mentioned traffic demand prediction method and system. For example, when the processor executes the computer program, it implements the method process and related operations such as steps S101 to S103, steps S201 to S203, steps S301 to S303 and steps S401 to S403.

[0135] The embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0136] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application is described in detail with reference to the above embodiments, a person skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting traffic demand, characterized in that: include: Acquire map data, perform grid generation processing on the map data and extract geographic grid features, the map data including one or more of point map data, line map data and surface map data, and the geographic grid features including feature vectors describing multiple geographic indicators of each grid space; Acquire first traffic data, and decompose the first traffic data according to the geographic grid feature to obtain second traffic data; Spatial traffic data and time series traffic data are generated according to the second traffic data and the geographic grid features. The spatial traffic data includes data of traffic demand prediction of gridded spatial distribution, and the time series traffic data includes data of traffic demand prediction of normalized time series distribution.

2. The method according to claim 1, characterized in that The process of performing raster generation processing on the map data and extracting geographic raster features comprises: generating a grid network according to data anchor points, the data anchor points including fine-grained traffic data as gridded anchor points; The geographic grid features of the map data are extracted according to the grid network, wherein the geographic grid features include one or more of the point grid features, the line grid features and the surface grid features.

3. The method according to claim 2, characterized in that The step of extracting geographic grid features of the rasterized map data comprises: In the case where the map data includes the point map data, point grid features of the point map data are extracted, and the point map data includes a plurality of different types of points of interest (POIs) distributed in the map space and a first geographical indicator of the point of interest, wherein the first geographical indicator includes altitude, longitude and latitude, and type of the point of interest.

4. The method according to claim 3, characterized in that The step of extracting the point grid features of the point map data comprises: The number of points of interest in each grid of the point-like map data is counted, and one or more numbers of points of interest corresponding to each grid are combined into a point-like grid feature.

5. The method according to claim 2, characterized in that: The extracting of geographic grid features of the map data according to the grid network also includes: In the case where the map data includes the linear map data, linear grid features of the linear map data are extracted, the linear map data includes road network data distributed in the map space and second geographical indicators of the road network data, the second geographical indicators include road length, road direction and road geometry.

6. The method according to claim 5, characterized in that The step of extracting the linear grid features of the linear map data comprises: The shortest distance between the road inside each grid and the center point of the grid is calculated, and one or more of the shortest distances corresponding to each grid are combined into a linear grid feature.

7. The method according to claim 2, characterized in that The extracting of geographic grid features of the map data according to the grid network also includes: In the case where the map data includes the planar map data, the planar grid features of the planar map data are extracted, the planar map data includes a plurality of geometric figures describing building contours distributed in the map space and third geographical indicators of the geometric figures, the third geographical indicators including population size, building area, land use type and building purpose.

8. The method according to claim 7, characterized in that The step of extracting the planar grid features of the planar map data comprises: The area ratio of the planar map data in each grid is counted, and one or more area ratios corresponding to each grid are combined into a planar grid feature.

9. The method according to claim 2, characterized in that: Also includes: In the case where the extracted geographic grid features include multiple ones of the point grid features, the linear grid features and the surface grid features, multiple ones of the point grid features, the linear grid features and the surface grid features are combined to obtain a grid combination feature, the grid combination feature includes geographic indicators that describe multiple types of the map data in each grid space, and the geographic indicators include the first geographic indicator, the second geographic indicator or the third geographic indicator.

10. The method according to claim 9, characterized in that The step of combining multiple types of the point grid features, the line grid features, and the surface grid features to obtain grid combination features includes: The grid combination feature is obtained by aligning and unifying multiple types of the point grid features, the line grid features and the surface grid features, wherein the alignment and unification operation includes unification of coordinate reference system, unification of grid coverage, unification of grid resolution and unification of grid azimuth, and the geographic grid features include the grid combination feature.

11. The method according to any one of claims 1 to 10, characterized in that: Decomposing the first traffic data according to the geographic grid features to obtain second traffic data includes: Decomposing the first traffic data from different base stations or cells according to a base station coverage model to obtain a plurality of cell traffic flows, wherein the cell traffic flows include the first traffic data in a cell unit; Decomposing each of the cell traffic flows into a plurality of groups of cell grid flows according to the geographic grid features according to the traffic decomposition model, wherein the cell grid flows include a fine-grained distribution pattern of the cell traffic flows that have been subjected to gridding; The plurality of groups of cell grid traffic are combined and aggregated to obtain the second traffic data.

12. The method according to claim 11, characterized in that The first traffic data includes time-series traffic flows from a plurality of base stations, and the second traffic data includes the time-series traffic flows distributed in a grid.

13. The method according to claim 12, characterized in that The first traffic data also includes one or more of the industrial parameter data and the user experience data; the second traffic data also includes one or more of the industrial parameter data and the user experience data in a grid-based distribution.

14. The method according to any one of claims 1 to 13, characterized in that: The generating of spatial traffic data and temporal traffic data according to the second traffic data and the geographic grid features comprises: Inputting the second traffic data and the geographic grid features into a spatial generation model to generate spatial traffic data, wherein the spatial traffic data includes a gridded distribution value of traffic demand prediction in a spatial domain; Input multiple groups of clustered traffic flows into a timing generation model to generate multiple groups of timing traffic data, wherein the timing traffic data include multiple groups of normalized distribution patterns of traffic demand prediction in the time domain, and the multiple groups of clustered traffic flows include traffic data classified according to cluster discrimination results for the second traffic data.

15. The method according to claim 14, wherein the step of inputting multiple groups of clustered traffic flows into a time series generation model to generate multiple groups of time series traffic data comprises: Performing cluster analysis on the second traffic data according to the geographical indicator type to obtain a cluster identification result, and classifying the second traffic data according to the cluster identification result to obtain multiple groups of clustered traffic flows, wherein the geographical indicator type includes multiple types of geographical indicators extracted from the geographical grid features; The multiple groups of clustered traffic flows are respectively input into a time series generation model to generate multiple groups of time series traffic data, and the time series generation model includes an adversarial generation network model.

16. The method according to claim 14, characterized in that The method further comprises: The spatial traffic data and the multiple groups of time-series traffic data are grid-aligned and aggregated to obtain spatiotemporal traffic data, wherein the spatiotemporal traffic data includes data of multi-mode traffic demand prediction decoupled in spatiotemporal domains.

17. A traffic demand prediction device, characterized in that: include: A data processing and extraction module, used to obtain map data, perform grid generation processing on the map data and extract geographic grid features, wherein the map data includes one or more of point map data, line map data and surface map data, and the geographic grid features include feature vectors describing multiple geographic indicators of each grid space; A traffic space decomposition module, used for acquiring first traffic data, and decomposing the first traffic data according to the geographic grid features to obtain second traffic data; A spatiotemporal data generation module is used to generate spatial traffic data and time series traffic data based on the second traffic data and the geographic grid features, wherein the spatial traffic data includes data for predicting traffic demand for gridded spatial distribution, and the time series traffic data includes data for predicting traffic demand for normalized time series distribution.

18. The device according to claim 17, characterized in that The data processing and extraction module is specifically used for: generating a grid network according to data anchor points, the data anchor points including fine-grained traffic data as gridded anchor points; The geographic grid features of the map data are extracted according to the grid network, wherein the geographic grid features include one or more of the point grid features, the line grid features and the surface grid features.

19. The device according to claim 18, characterized in that The data processing and extraction module extracts geographic grid features of the map data according to the grid network, including: In the case where the map data includes the point map data, extracting point grid features of the point map data, the point map data includes a plurality of different types of points of interest (POIs) distributed in a map space and a first geographical indicator of the point of interest, the first geographical indicator including altitude, longitude and latitude, and type of the point of interest; Wherein, the step of extracting the point grid features of the point map data comprises: The number of points of interest in each grid of the point-like map data is counted, and one or more numbers of points of interest corresponding to each grid are combined into a point-like grid feature.

20. The device according to claim 18, characterized in that The data processing and extraction module extracts geographic grid features of the map data according to the grid network, and further includes: In the case where the map data includes the linear map data, extracting linear grid features of the linear map data, the linear map data including road network data distributed in a map space and a second geographical indicator of the road network data, the second geographical indicator including road length, road direction and road geometry; Wherein, the extracting of linear grid features of the linear map data comprises: The shortest distance between the road inside each grid and the center point of the grid is calculated, and one or more of the shortest distances corresponding to each grid are combined into a linear grid feature.

21. The device according to claim 18, characterized in that The data processing and extraction module extracts geographic grid features of the map data according to the grid network, and further includes: In the case where the map data includes the planar map data, extracting planar grid features of the planar map data, the planar map data including a plurality of geometric figures describing building outlines distributed in a map space and third geographical indicators of the geometric figures, the third geographical indicators including population, building area, land use type, and building purpose; Wherein, the step of extracting the planar grid features of the planar map data comprises: The area ratio of the planar map data in each grid is counted, and one or more area ratios corresponding to each grid are combined into a planar grid feature.

22. The device according to claim 18, characterized in that The data processing and extraction module is also used to: In the case where the extracted geographic grid features include multiple ones of the point grid features, the linear grid features and the surface grid features, multiple ones of the point grid features, the linear grid features and the surface grid features are combined to obtain a grid combination feature, the grid combination feature includes geographic indicators that describe multiple types of the map data in each grid space, and the geographic indicators include the first geographic indicator, the second geographic indicator or the third geographic indicator.

23. The device according to claim 22, characterized in that The step of combining multiple types of the point grid features, the line grid features, and the surface grid features to obtain grid combination features includes: The grid combination feature is obtained by aligning and unifying multiple types of the shape grid feature, the linear grid feature and the surface grid feature, wherein the alignment and unification operation includes unification of coordinate reference system, unification of grid coverage, unification of grid resolution and unification of grid azimuth, and the geographic grid feature includes the grid combination feature.

24. The device according to any one of claims 15 to 23, characterized in that The traffic space decomposition module is specifically used for: Decomposing the first traffic data from different base stations or cells according to a base station coverage model to obtain a plurality of cell traffic flows, wherein the cell traffic flows include the first traffic data in a cell unit; Decomposing each of the cell traffic flows into a plurality of groups of cell grid flows according to the geographic grid features according to the traffic decomposition model, wherein the cell grid flows include a fine-grained distribution pattern of the cell traffic flows that have been subjected to gridding; The plurality of groups of cell grid traffic are combined and aggregated to obtain the second traffic data.

25. The device according to claim 24, characterized in that The first traffic data includes time-series traffic flows from a plurality of base stations, and the second traffic data includes the time-series traffic flows distributed in a grid.

26. The device according to any one of claims 17 to 25, characterized in that The spatiotemporal data generation module is specifically used for: Inputting the second traffic data and the geographic grid features into a spatial generation model to generate spatial traffic data, wherein the spatial traffic data includes a gridded distribution value of traffic demand prediction in a spatial domain; Input multiple groups of clustered traffic flows into a timing generation model to generate multiple groups of timing traffic data, wherein the timing traffic data include multiple groups of normalized distribution patterns of traffic demand prediction in the time domain, and the multiple groups of clustered traffic flows include traffic data classified according to cluster discrimination results for the second traffic data.

27. The apparatus according to claim 26, wherein the step of inputting multiple groups of clustered traffic flows into a time series generation model to generate multiple groups of time series traffic data comprises: Performing cluster analysis on the second traffic data according to the geographical indicator type to obtain a cluster identification result, and classifying the second traffic data according to the cluster identification result to obtain multiple groups of clustered traffic flows, wherein the geographical indicator type includes multiple types of geographical indicators extracted from the geographical grid features; The multiple groups of clustered traffic flows are respectively input into a time series generation model to generate multiple groups of time series traffic data, and the time series generation model includes an adversarial generation network model.

28. The device according to claim 26, characterized in that The spatiotemporal data generation module is further used for: The spatial traffic data and the multiple groups of time-series traffic data are grid-aligned and aggregated to obtain spatiotemporal traffic data, wherein the spatiotemporal traffic data includes data of multi-mode traffic demand prediction decoupled in spatiotemporal domains.

29. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores instructions, and when the processor executes the instructions in the memory, the method according to any one of claims 1 to 16 can be implemented.

30. A non-volatile storage medium, characterized in that: The method comprises a plurality of instructions, and when the instructions are executed by an electronic device, the method according to any one of claims 1 to 16 can be implemented.

Images