5G indoor distribution smart portrait construction method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of mobile communication network optimization, in particular to a 5G indoor distribution smart portrait construction method and system based on multi-source data fusion, and constructs a building value, coverage and flow three-dimensional smart portrait model by fusing user-level operation analysis data, MDT data, geographical rasterization data and building attribute data. And precise planning and optimization of 5G indoor distribution construction are realized through a comprehensive evaluation model, intelligent analysis is realized through multi-source data fusion and AI modeling, manual intervention is reduced, the production efficiency of indoor distribution construction is improved, high-value and high-traffic buildings are configured preferentially, the traffic of a single cell is greatly improved, optimal release of network resources is realized, and the method is suitable for popularization and application. According to the invention, the problems of insufficient precision, single dimension and low efficiency in traditional indoor coverage evaluation are solved, optimal network resource configuration taking users as the center is realized, and the 5G indoor coverage quality and the operation efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mobile communication network optimization, in particular to a 5G room distribution wisdom portrait construction method and system based on multi-source data fusion. BACKGROUND

[0002] With the advancement of 5G network commercialization process, 70% of mobile communication traffic is concentrated in indoor environment, and indoor coverage becomes the core link of network optimization. Generally, the indoor mobile communication environment is improved by using a room distribution system, that is, the radio frequency signal of the base station is transmitted to each node in the indoor environment to complete the mobile signal distribution coverage of the indoor scene, so that the indoor mobile terminal user can also enjoy the high-speed mobile signal network.

[0003] The existing 5G room distribution construction evaluation method has the following limitations: (1) it depends on the macro station coverage to predict the indoor effect, and needs to test the wall loss on site, which is low in efficiency; (2) it evaluates the building value based on the cell dimension, and is not accurate to the user level, resulting in inaccurate value identification; (3) the data sources are scattered and the analysis relies on manual work, which is difficult to realize intelligent decision-making, resulting in the problems of insufficient precision, single dimension, low efficiency and weak intelligence in the existing 5G room distribution evaluation. Therefore, a 5G room distribution wisdom portrait construction method and system based on multi-source data fusion are proposed. SUMMARY

[0004] In order to solve the technical problems existing in the prior art, the present application provides a 5G room distribution wisdom portrait construction method and system based on multi-source data fusion.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a 5G room distribution wisdom portrait construction method based on multi-source data fusion, which specifically comprises the following steps:

[0006] S1, data acquisition and preprocessing: obtaining user-level operation analysis data, 4G / 5G MDT data, geographic gridding data, building attribute data and cell-level traffic data, and performing data cleaning and association;

[0007] S2, building value wisdom portrait modeling: calculating the building user value based on the resident user operation analysis data, combining the building coverage scene characteristics to calculate the scene value, and comprehensively obtaining the building value score and grade;

[0008] S3, building coverage wisdom portrait modeling: using 5G penetration loss model and 4G / 5G coverage difference model, and evaluating the 5G macro station coverage quality in the building based on MDT data;

[0009] S4, building traffic wisdom portrait modeling: through building and room distribution cell (indoor distribution system cell) traffic association analysis, realizing building traffic grading and high backflow state identification;

[0010] S5, comprehensive image evaluation: based on building value, coverage and traffic data, 5G room division to be built pool evaluation rules are constructed, and construction priority and scheme suggestions are output.

[0011] Preferably, in the step S1, the obtained user-level operation analysis data includes 5G user identification, VIP level and ARPU value; 4G / 5G MDT number includes RSRP and latitude and longitude; geographical gridding data includes application scene and building structure; building attribute data includes building area and building purpose; and cell-level traffic data is 4G or 5G room division cell traffic;

[0012] In data cleaning, obviously unreasonable outliers are removed, and data format and time granularity are unified.

[0013] Preferably, in the step S2, the building user value is scored by 5G terminal user number, ARPU average value and VIP user number respectively, and the weight proportions of the three are 0.4, 0.4 and 0.2 respectively, wherein:

[0014]

[0015]

[0016] The VIP user score is divided into 0, 10 or 20 points according to the comparison between the user number and the single cell average value of the whole network.

[0017] Preferably, in the step S3, 5G penetration loss model and 4G / 5G coverage difference model are first constructed, then 5G MDT data-limited buildings are screened out, RSRP of each 5G frequency band cell in the corresponding base station coverage in each surrounding grid is estimated through the 4G / 5G coverage difference model; and RSRP in the building is predicted through the 5G penetration loss model, so as to evaluate the 5G macro station coverage quality in the building.

[0018] Preferably, in the step S3, when constructing the 5G penetration loss model, the collected 4G MDT and 5G MDT data are respectively divided into grids with latitude and longitude as lines to form a grid, the number of sampling points and the average signal receiving power of 4G and 5G cells inside each grid are calculated, and the building layer is layered to obtain the building data of each layer. The grid correlation technology is used to identify the grid with a sampling point number greater than 100 as a data sufficient grid, and when the proportion of data sufficient grids in each individual building exceeds 30%, the building is marked as a data sufficient building, so as to screen out buildings with sufficient 5G MDT data.

[0019] The sectors corresponding to the top three buildings in the number of internal sampling points are selected from the data sufficient buildings, and are summarized to determine the base stations corresponding to the three sectors.

[0020] The number of sampling points and the average signal receiving power in the top three buildings and the two surrounding grating rings are extracted synchronously. Then, the 5G engineering parameter data of the sector are associated with the corresponding base station, and the latitude, longitude, frequency band, antenna height and direction angle information of the base station are obtained to calculate the geographical relationship between the base station and the building, including five dimensional features of base station height, distance, incident angle, frequency and obstacle layer number. After obtaining the five dimensional features and the average signal receiving power in the building, a 5G penetration loss model is constructed by linear regression model training. After obtaining the geographical relationship between the base station and the building, the signal receiving power in the building is evaluated by the 5G penetration loss model.

[0021] Preferably, in step S3, when constructing the 4G / 5G outdoor coverage difference model, based on the grid divided when constructing the penetration loss model, 4G / 5G MDT data covering different dates are used to preliminarily screen the grid. The screening condition is that the total number of sampling points exceeds 100 and the data information is rich. The signal coverage details are associated with the preliminarily screened grid. For each grid after preliminary screening, the cells covered by 4G and 5G signals are selected again, and the average signal strength of the 4G and 5G signals of the twice screened cells is calculated respectively.

[0022] Combined with the 4G and 5G engineering parameter data, the geographical coordinates, frequency band and antenna height information of the twice screened cells are obtained, the distance, incident angle and obstacle layer number after twice screening are calculated, and the data is classified according to the frequency band.

[0023] In each frequency band combination group, based on the four features of base station height, distance, incident angle and obstacle layer number, the four dimensional features and the signal strength index in the building are obtained, and then the linear regression model is trained to construct the 4G / 5G coverage difference model. After obtaining the geographical relationship between the base station and the building, the average signal strength of each frequency band in the building is evaluated by the 4G / 5G coverage difference model.

[0024] Preferably, in step S4, the building traffic classification includes: the daily average of the room partition cell traffic is high traffic threshold 1, the daily average of the building user traffic is high traffic threshold 2, and according to the size relationship between the building traffic and the daily average of the room partition cell traffic and the daily average of the building user traffic, the building traffic is divided into four levels of extremely high, high, medium and low.

[0025] In the area of 4G and 5G network common coverage, part of mobile data traffic of 5G users who have opened 5G service is not transmitted through 5G network, but is flowed back to 4G network for transmission, the traffic of 5G users flowed back to 4G cell is counted, and the cell whose daily average flow back is greater than or equal to 10GB is marked as high flow back, and finally the building traffic evaluation results including time, building name, building number, building traffic level and whether it is high flow back are obtained.

[0026] After marking, the 5G high flow back business optimization is performed: the same address mapping relationship between 4G and 5G indoor distribution cells is established based on the relationship between 5G indoor coverage quality and same address mapping, and the 5G coverage quality analysis of the indoor scene of the 4G distribution cell which has been constructed is accurately realized based on 5G user flow back and 5G occupation coverage statistics, thereby supporting the daily optimization of 5G high flow back area.

[0027] Preferably, in the step S5, the 5G distribution cell to be built pool evaluation rule includes: high value buildings are directly included in the to-be-built pool; in general value buildings, buildings with extremely high traffic, high traffic or medium traffic are included in the to-be-built pool.

[0028] In combination with 4G distribution base data, whether a building is included in the 5G distribution to-be-built pool is judged according to the 5G distribution to-be-built pool rule, and when 5G distribution construction is hindered by cost, resident obstruction and the like, a co-site macro station optimization scheme is adopted, and the scheme suggestion of "newly built 5G distribution" or "co-site macro station optimization" is output.

[0029] A 5G distribution intelligent portrait construction system based on multi-source data fusion, the construction system includes a data acquisition module, a value portrait module, a coverage portrait module, a traffic portrait module and a comprehensive evaluation module; the data acquisition module is used for integrating multi-source heterogeneous data; the value portrait module performs building user value calculation; the coverage portrait module performs 5G penetration model and coverage evaluation; the traffic portrait module performs traffic classification; and the comprehensive evaluation module performs to-be-built pool evaluation.

[0030] The construction system further includes a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer readable storage medium is executed by a processor to realize the steps of the construction method.

[0031] Preferably, the data sources supported by the data acquisition module include user-level distribution data of a BSS system, MDT measurement data of a network management system, geographic grid data of a GIS system and building attribute data of a CRM system.

[0032] Compared with the prior art, the present application has the following beneficial effects:

[0033] 1. The application fuses user-level operation analysis data, MDT data, geographic gridded data and building attribute data, constructs a building value, coverage, traffic three-dimensional intelligent portrait model, combines user quantity, ARPU, coverage scene and area analysis, realizes accurate planning and optimization of 5G indoor distribution construction through comprehensive evaluation model, solves the problems of insufficient precision, single dimension and low efficiency in traditional indoor coverage evaluation, realizes optimal configuration of network resources based on users, improves 5G indoor coverage quality and operation efficiency;

[0034] 2. The application upgrades the evaluation difficulty from the cell dimension to the user dimension, combines MDT data to replace field testing, improves the evaluation accuracy, realizes intelligent analysis through multi-source data fusion and AI modeling, reduces manual intervention, improves the production efficiency of indoor distribution construction, preferentially configures high-value and high-traffic buildings, greatly improves the single-cell traffic, realizes optimal network resource allocation, constructs a user-centered coverage model, effectively solves the problem of deep coverage, and improves user satisfaction. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 A process schematic diagram of the 5G indoor distribution intelligent portrait system based on multi-fusion of the application is established;

[0036] Figure 2 A schematic diagram of the existing building-level 5G coverage portrait scheme of the application is established;

[0037] Figure 3 A building value intelligent portrait modeling algorithm flowchart of the application is established;

[0038] Figure 4 A scene value table of the application is established;

[0039] Figure 5 A 5G penetration loss model and outdoor coverage difference flowchart of the application is established;

[0040] Figure 6 A building traffic intelligent portrait algorithm modeling flowchart of the application is established;

[0041] Figure 7 A building traffic value evaluation representation of the application is established;

[0042] Figure 8 A building whether to be included in the 5G indoor distribution to be built pool rule representation of the application is established;

[0043] Figure 9 A cell distribution schematic diagram in embodiment two of the application is established;

[0044] Figure 10 A penetration loss model weight reference table of the application is established. DETAILED DESCRIPTION

[0045] The present invention will be further described below with reference to the accompanying drawings and embodiments, which illustrate the above and other technical features and advantages of the present invention. However, the following embodiments are merely preferred embodiments of the present invention and are not exhaustive.

[0046] Example 1:

[0047] like Figures 1-10 As shown, this invention provides a method for constructing a 5G indoor distributed smart profile based on multi-source data fusion. The method specifically includes the following steps:

[0048] S1, Data Acquisition and Preprocessing: Acquire user-level business analysis data, 4G / 5G MDT data, geographic rasterized data, building attribute data, and community-level traffic data, and perform data cleaning and correlation;

[0049] MDT data is network performance measurement data containing location information that is automatically collected using user terminals. It is a widely adopted method for operators to measure signals.

[0050] S2, Building Value Intelligent Profile Modeling: Calculate the building user value based on the resident user operation analysis data, calculate the scene value by combining the characteristics of the building coverage scene, and obtain the building value score and level in a comprehensive manner;

[0051] S3, Smart Building Coverage Profile Modeling: Using 5G penetration model and 4G / 5G coverage difference model, the coverage quality of 5G macro base stations in buildings is evaluated based on MDT data;

[0052] S4, Smart Building Traffic Profile Modeling: Through traffic correlation analysis between buildings and indoor distribution system (IDS) communities, the system realizes the classification of building traffic and the identification of high backflow conditions;

[0053] S5, Comprehensive Profile Assessment: Based on building value, coverage and traffic data, construct assessment rules for the 5G indoor distribution system pool to be built, and output construction priorities and solution suggestions.

[0054] In this embodiment, in step S1, the acquired user-level business analysis data includes 5G user identifiers, VIP levels, and ARPU (Average Revenue Per User); 4G / 5G MDT data includes RSRP (i.e., the average power of the base station reference signal received by the user terminal, which directly reflects the signal propagation quality between the terminal and the base station), latitude and longitude, station name, latitude and longitude, RSRP, base station cell PCI, and frequency band; geographic rasterization data includes application scenarios and building structures; building attribute data includes building area and building usage; and cell-level traffic data is 4G or 5G indoor distributed cell traffic.

[0055] During data cleaning, allowable indicator ranges, units, or requirements are set as standards to eliminate obviously unreasonable outliers, such as latitude and longitude, hanging height, RSRP type and communication professional data. Data association is performed by matching users and buildings using latitude and longitude, that is, matching the distance between the user and the building location using latitude and longitude. For example, if the two are within 100 meters in the city center and 200 meters in the county town, it is considered a match. The data format and time granularity are also unified.

[0056] In this embodiment, in step S2, considering actual needs and investment value, in order to accurately allocate limited resources according to needs, the building user value is scored by the number of 5G terminal users, the average ARPU (i.e., the communication service revenue contributed by each user), and the number of VIP users. The more VIP users, the higher the score. The weights of the three are 0.4, 0.4, and 0.2 respectively.

[0057]

[0058]

[0059] VIP user scores are categorized as 0, 10, or 20 points based on the comparison between the number of VIP users and the average score for a single cell across the entire network.

[0060] In this embodiment, in step S3, a 5G penetration loss model and a 4G / 5G coverage difference model are first constructed. Then, buildings with limited 5G MDT data are selected. The RSRP of each 5G frequency band cell covered by the corresponding base station in each surrounding grid is estimated by the 4G / 5G coverage difference model. Then, the RSRP in the building is predicted by the 5G penetration loss model to evaluate the 5G macro base station coverage quality in the building.

[0061] Wherein: the penetration loss value is the difference between the average power of the signal emitted by the base station and the average power of the signal received inside the building;

[0062] Coverage difference is the difference between the average signal strength of 4G cells and the average signal strength of 5G cells.

[0063] In this embodiment, in step S3, when constructing the 5G penetration loss model, the collected 4G MDT and 5G MDT data are divided into grids using latitude and longitude as lines to form a grid. The number of sampling points of 4G and 5G cells and the average signal received power of each grid are calculated. At the same time, the building layer is layered to obtain the building data of each layer. Using grid association technology, grids with more than 100 sampling points are marked as grids with sufficient data. In each individual building, when the proportion of grids with sufficient data exceeds 30%, the building is marked as a building with sufficient data, thereby filtering out buildings with sufficient 5G MDT data.

[0064] From buildings with sufficient data, select the sectors corresponding to the top three buildings with the most internal sampling points (excluding other data), and summarize them to determine the base stations corresponding to the three sectors. Usually, each base station consists of three sectors, with the base station as the center and the surrounding 360° divided into three sectors.

[0065] Simultaneously extract the number of sampling points and the average signal received power within the top three buildings and the two outer rings of grids surrounding the buildings; then associate the 5G engineering parameter data of the sector with the corresponding base station, and obtain the base station's latitude, longitude, frequency band, antenna height, and azimuth information to calculate the geographical relationship between the base station and the building. Specifically, this includes five dimensions: base station height, distance (distance from the base station to the building center), incident angle (the angle between the base station's straight-line angle and the building's azimuth angle), frequency, and number of obstacle layers (i.e., the number of other buildings the signal passes through in a straight line). After obtaining the five dimensions and the average signal received power within the building, select a sufficient number (e.g., 100) residential communities and train them using a linear regression model to construct a 5G penetration loss model. After obtaining the geographical relationship between the base station and the building, evaluate the signal received power within the building using the 5G penetration loss model. The specific calculation formula for the 5G penetration loss model is as follows:

[0066]

[0067] Where R is the independent variable (explanatory variable), used to explain 85% of the variation in the dependent variable (predicted variable), and the weights of each term are as follows: Figure 10 For example, if an existing base station is 220m away from the target coverage area, the base station antenna is mounted at a height of 38 meters, the frequency band is 1800MHz, the number of obstruction layers is 1, and the angle of incidence is 4 degrees, then the penetration loss value = 4 + 220. 0.12+1800 0.025 + 5 - 2 + 5 = 83.4 dB.

[0068] In this embodiment, in step S3, when constructing the 4G / 5G outdoor coverage difference model, based on the grids divided when constructing the penetration damage model, 4G / 5G MDT data covering different dates is used to initially screen the grids. The screening criteria are that the total number of sampling points exceeds 100 and the data information is rich, that is, the 4G / 5G MDT data is complete, including latitude and longitude, tower height, azimuth angle, and frequency band. The signal coverage details are associated with the initially screened grids. For each grid after the initial screening, cells that are simultaneously covered by 4G and 5G signals are screened again. The average signal strength of 4G and 5G signals of the cells selected in the second screening is calculated respectively. The range of the average signal strength is usually within ;

[0069] By combining 4G and 5G engineering parameter data, the geographic coordinates, frequency bands, and antenna height information of the cells after secondary screening are obtained. The distance, incident angle, and number of obstacle layers after secondary screening are calculated. The data are then classified according to frequency band, such as 4G frequency band and 5G frequency band, so as to control variables during evaluation.

[0070] Within each frequency band combination, based on four features—base station height, distance, incident angle, and number of obstacle layers—four-dimensional features and signal strength indicators within the building are obtained. Then, a 4G / 5G coverage difference model is constructed by training a linear regression model. After obtaining the geographical relationship between the base station and the building, the average signal strength of each frequency band within the building is evaluated through the 4G / 5G coverage difference model.

[0071] In this embodiment, in step S4, the daily average traffic flow of the indoor distributed antenna system (DAS) is set to the high traffic flow threshold 1, and the daily average traffic flow of the building users is set to the high traffic flow threshold 2. Based on the relationship between the building traffic flow and the daily average traffic flow of the indoor DAS and the daily average traffic flow of the building users, the building traffic flow is divided into four levels: extremely high, high, medium, and low. Details of the classification can be found in [link to relevant documentation]. Figure 7 ;

[0072] In areas covered by both 4G and 5G networks, some mobile data traffic of 5G users who have activated 5G services is not transmitted through the 5G network, but instead flows back to the 4G network. The traffic that flows back from 5G users to 4G cells is counted. A single cell with an average daily backflow of 10GB or more is marked as high backflow. The final result includes: time, building name, building number, building traffic level, and whether it is a high backflow building.

[0073] After marking, the co-location mapping relationship between 4G and 5G indoor distributed cells is established based on the 5G indoor coverage quality relationship of co-location mapping. Then, based on 5G user backflow and 5G coverage occupancy statistics, the 5G coverage quality analysis of indoor scenarios where 4G indoor distributed cells have been built can be accurately realized, supporting daily optimization of 5G high backflow areas.

[0074] Co-location mapping refers to different network resources, data, or configurations within the same physical location (such as the same building or the coverage area of ​​the same base station).

[0075] The daily average traffic volume of an indoor distributed antenna system (DAS) cell is the daily average traffic volume of the indoor portion of the filtered access cell-level traffic data.

[0076] The average daily traffic volume of a building user is calculated by extracting traffic data generated by all users in the building through network management, summarizing the data by building ID, based on information on resident users and user business analysis data.

[0077] In this embodiment, in step S5, the evaluation rules for the 5G indoor distribution network (UDN) pool are as follows: high-value buildings are directly included in the pool; among general-value buildings, those with extremely high traffic, high traffic (medium / poor coverage), or medium traffic (800M poor coverage) are included in the pool; at the same time, priority is marked in the pool, with high-traffic buildings having a priority of 1 and medium-traffic buildings having a priority of 2. If there are network complaints or owner network needs in medium-traffic buildings in the future, the priority can be increased to 1.

[0078] Based on the 4G indoor distribution data and the rules for including buildings in the 5G indoor distribution pool, it is determined whether a building should be included in the 5G indoor distribution pool. If the construction of 5G indoor distribution is hindered by cost issues or resident obstruction, a co-site macro base station optimization scheme is adopted, and a scheme suggestion of "building a new 5G indoor distribution" or "co-site macro base station optimization" is output.

[0079] A 5G indoor distributed antenna system (DAS) for building smart profiles based on multi-source data fusion includes a data acquisition module, a value profile module, a coverage profile module, a traffic profile module, and a comprehensive evaluation module. The data acquisition module integrates heterogeneous data from multiple sources. The value profile module performs building user value calculation. The coverage profile module performs 5G penetration modeling and coverage evaluation. The traffic profile module performs traffic classification. The comprehensive evaluation module combines the data acquisition module, the value profile module, the coverage profile module, and the traffic profile module to perform evaluation of the target database.

[0080] The system also includes a computer-readable storage medium that stores a computer program that, when executed by a processor, implements the steps of the construction method.

[0081] In this embodiment, the data acquisition module supports data sources including user-level economic analysis data from the BSS system (Business Support System), MDT measurement data from the network management system, geographic raster data from the GIS system (Geographic Information System), and building attribute data from the CRM system (Customer Relationship System).

[0082] Example 2:

[0083] Taking communities A, B, C, and D as examples, the specific operation process is as follows:

[0084] S1: Data Acquisition and Preprocessing: Acquire the following data from cells A, B, C, and D:

[0085] (1) Number of 5G terminal users, average ARPU (i.e., the revenue from communication services contributed by each user), and number of VIP users;

[0086] (2) 4G / 5G MDT data (signal strength indicators);

[0087] (3) Geographic raster data and building attribute data;

[0088] (4) Daily average traffic data at the community level;

[0089] Based on the operator's engineering parameter data, there are 5 existing outdoor base stations around the community. Figure 9 The five base stations marked in red are outdoor stations 1, 2, 3, 4, and 5. The signal strength and traffic data are taken from these five base stations.

[0090] S2: Smart Building Value Profile Modeling: Based on the number of 5G terminal users, average ARPU, and number of VIP users, scores are calculated respectively to obtain the scores of communities A, B, C, and D as A[40, 48, 20], B[20, 60, 10], C[32, 60, 10], and D[20, 32, 0]. Therefore, communities A, B, and C are all high-value buildings, and D is a low-value building.

[0091] S3, Smart Building Coverage Modeling: The threshold values ​​of cells A, B, C, and D are estimated to be [15dB, 22dB, 18dB, 12dB] and [13dB, 22dB, 15dB, 10dB] respectively, based on the 5G penetration model and the 4G / 5G coverage difference model.

[0092] S4, Smart Building Traffic Profile Modeling: The comparison shows that the building traffic levels of the four communities are [high, high, medium, low].

[0093] S5, Comprehensive Profile Assessment: Assess the 5G indoor distribution pool of buildings to be built, and output building coverage plan suggestions based on the assessment results. Combined with the data above, it can be concluded that the buildings A, B, C and D can be included in the pool of buildings to be built.

[0094] The above are merely preferred embodiments of the present invention and are illustrative in nature, not restrictive. Those skilled in the art will understand that many changes, modifications, and even equivalents can be made within the spirit and scope defined by the claims of the present invention, all of which will fall within the protection scope of the present invention.

Claims

1. A 5G room intelligent image construction method based on multi-source data fusion, characterized in that, The construction method specifically comprises the following steps: S1, data acquisition and preprocessing: obtaining user-level operation analysis data, 4G / 5G MDT data, geographic gridding data, building attribute data and cell-level traffic data, and performing data cleaning and association; S2, building value wisdom portrait modeling: calculating building user value based on resident user operation analysis data, calculating scene value in combination with building coverage scene characteristics, and comprehensively obtaining building value score and grade; S3, building coverage wisdom portrait modeling: evaluating 5G macro station coverage quality in the building based on MDT data by using a 5G penetration model and a 4G / 5G coverage difference model; S4, building traffic wisdom portrait modeling: realizing building traffic grading and high backflow state recognition through building and indoor distribution system (IDC) cell traffic association analysis; S5, comprehensive portrait evaluation: based on building value, coverage and traffic data, constructing a 5G IDC construction pool evaluation rule, and outputting construction priority and scheme suggestion. 2.The 5G room intelligence image construction method based on multi-source data fusion according to claim 1, wherein, In the step S1, the obtained user-level operation analysis data comprises 5G user identification, VIP level and ARPU value; the 4G / 5G MDT data comprises RSRP and longitude and latitude; the geographic gridding data comprises application scene and building structure; the building attribute data comprises building area and building purpose; and the cell-level traffic data is 4G or 5G IDC cell traffic; In data cleaning, obviously unreasonable outliers are removed, and data format and time granularity are unified. 3.The 5G room intelligence image construction method based on multi-source data fusion of claim 1, wherein, In the step S2, the building user value is scored by 5G terminal user number, ARPU average value and VIP user number, and the weight proportions of the three are 0.4, 0.4 and 0.2 respectively, wherein: The VIP user score is divided into 0, 10 or 20 points according to the comparison between the user number and the single cell average value in the whole network. 4.The 5G room intelligence image construction method based on multi-source data fusion of claim 1, wherein, In the step S3, a 5G penetration model and a 4G / 5G coverage difference model are first constructed, then buildings with limited 5G MDT data are selected, the RSRP of each 5G frequency band cell in the corresponding base station in each surrounding grid is estimated by the 4G / 5G coverage difference model, and then the RSRP in the building is predicted by the 5G penetration model, so as to evaluate the 5G macro station coverage quality in the building. 5.The 5G room intelligence image construction method based on multi-source data fusion according to claim 4, characterized in that, In the step S3, when constructing the 5G penetration model, the collected 4G MDT and 5G MDT data are respectively divided into grids with longitude and latitude as lines to form a grid, the sampling point number and signal receiving power average of 4G and 5G cells inside each grid are calculated, the building layer is layered to obtain building data of each layer, the grid with a sampling point number greater than 100 is identified as a grid with sufficient data, and when the proportion of data sufficient grids in each separate building exceeds 30%, the building is marked as a data sufficient building, so as to select buildings with sufficient 5G MDT data; The sectors corresponding to the buildings with the top three internal sampling point numbers are selected from the data sufficient buildings, and are summarized to determine the base stations corresponding to the three sectors. The number of sampling points and the average signal receiving power in the top three buildings and the two circles of the building periphery are extracted synchronously; then the 5G engineering parameter data of the sector are associated with the corresponding base station, and the latitude and longitude, frequency band, antenna height and direction angle information of the base station are obtained to calculate the geographical relationship between the base station and the building, including five dimensional features of base station height, distance, incident angle, frequency and obstacle layer number; after obtaining the five dimensional features and the average signal receiving power in the building, a 5G penetration loss model is constructed through linear regression model training; after obtaining the geographical relationship between the base station and the building, the signal receiving power in the building is evaluated through the 5G penetration loss model. 6.The 5G room intelligence image construction method based on multi-source data fusion according to claim 4, characterized in that, In step S3, when constructing the 4G / 5G outdoor coverage difference model, based on the grid divided when constructing the penetration loss model, 4G / 5G MDT data covering different dates are used to preliminarily screen the grid, and the screening condition is that the total number of sampling points is more than 100 and the data information is rich; the signal coverage details are associated with the preliminarily screened grid; for each grid after preliminary screening, the cells covered by 4G and 5G signals are selected again, and the average signal strength of the 4G and 5G signals of the cells selected twice is calculated respectively; Combined with the engineering parameter data of 4G and 5G, the geographical coordinates, frequency band and antenna height information of the cells selected twice are obtained, the distance, incident angle and obstacle layer number after secondary screening are calculated, and the data are classified according to the frequency band; In each frequency band combination group, based on the four features of base station height, distance, incident angle and obstacle layer number, after obtaining the four dimensional features and the signal strength index in the building, a 4G / 5G coverage difference model is constructed through linear regression model training; after obtaining the geographical relationship between the base station and the building, the average signal strength of each frequency band in the building is evaluated through the 4G / 5G coverage difference model. 7.The 5G room intelligence image construction method based on multi-source data fusion of claim 1, wherein, In step S4, the building traffic classification includes: the daily average of the room cell traffic is high traffic threshold 1, the daily average of the building user traffic is high traffic threshold 2, and according to the size relationship between the building traffic and the daily average of the room cell traffic and the daily average of the building user traffic, the building traffic is divided into four levels of extremely high, high, medium and low; In the area covered by 4G and 5G networks, part of the mobile data traffic of 5G users who have opened 5G service is not transmitted through 5G network, but flows back to 4G network for transmission; the traffic of 5G users flowing back to 4G cells is counted, and the daily average of single cell flow back greater than or equal to 10GB is marked as high flow back, and finally the building traffic evaluation results including time, building name, building number, building traffic level and whether it is high flow back are obtained; After marking, the 5G high flow back business is optimized: the same address mapping relationship between 4G and 5G room cells is established based on the 5G indoor coverage quality relationship of the same address mapping, and the 5G coverage quality analysis of the 4G room which has been constructed in the indoor scene is accurately realized based on the 5G user flow back and 5G occupied coverage statistics, supporting the daily optimization of 5G high flow back area. 8.The 5G room intelligence image construction method based on multi-source data fusion according to claim 1, wherein, In the step S5, the 5G room division to-be-built pool evaluation rule includes: high-value buildings are directly included in the to-be-built pool; among general value buildings, buildings with extremely high traffic, high traffic or medium traffic are included in the to-be-built pool; In combination with 4G room division basic data, whether a building is included in the 5G room division to-be-built pool is judged according to the rule of including the 5G room division to-be-built pool, and when the 5G room division construction is hindered by cost and resident obstruction problems, a co-station macro station optimization scheme is adopted, and a scheme suggestion of "newly built 5G room division" or "co-station macro station optimization" is output.

9. A system for the 5G room intelligence image construction method based on multi-source data fusion according to any one of claims 1-8, characterized in that, The construction system includes a data acquisition module, a value portrait module, a coverage portrait module, a traffic portrait module and a comprehensive evaluation module; the data acquisition module is used for integrating multi-source heterogeneous data; the value portrait module performs building user value calculation; the coverage portrait module performs 5G penetration model and coverage evaluation; the traffic portrait module performs traffic classification; and the comprehensive evaluation module performs to-be-built pool evaluation. The construction system further includes a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer readable storage medium is executed by a processor to realize the steps of the construction method. 10.The 5G room intelligence image construction system based on multi-source data fusion of claim 9, wherein, The data acquisition module supports data sources including user-level divided data of a BSS system, MDT measurement data of a network management system, geographic grid data of a GIS system and building attribute data of a CRM system.

Images