Building occupancy rate data analysis method, apparatus and device, and storage medium

Abstract

The invention discloses a building occupancy rate data analysis method and device, equipment and a storage medium. The method comprises the steps of screening normal sampling point data for a fence area of a building, performing area sample expansion on the building and a surrounding area, increasing the number and range of sampling points, inputting multi-domain data capable of representing check-in characteristics in the sample expansion point data into a multi-domain data fusion model, and calculating the check-in rate of a target building; according to the embodiment of the invention, the priority of the user perception guarantee is determined according to the real estate occupancy rate, the perception guarantee automatic processing is realized, the working efficiency is improved, and the technical effects of reducing the cost and improving the efficiency are achieved.

Description

Technical Field

[0001] This application belongs to the technical field of wireless communication, and in particular relates to a method, apparatus, device and storage medium for analyzing building occupancy rate data. Background Technology

[0002] With the development of modern cities, the pace of urban construction and the establishment of communication infrastructure are changing rapidly. More and more wireless communication coverage is being deployed to improve the communication security of users within a region. However, there are very few technical solutions for accurately identifying the communication security needs of residents after they move into new housing developments.

[0003] Currently, the communication awareness assurance analysis for new residential buildings is based on the historical connection count of each base station in the base station list of the specified building to determine the resident users of the specified building; when a resident user is detected entering the specified building, the real-time signaling data of the resident user is obtained; and the network quality of the specified building is evaluated based on the real-time signaling data.

[0004] Current methods rely on user complaints, network metric anomalies triggering problem tickets, and information feedback from frontline staff. These methods are relatively slow in identifying the communication awareness and security needs of new buildings, and require frontline staff to follow up and collect information point by point, which is time-consuming and labor-intensive. Summary of the Invention

[0005] This application provides a method, apparatus, device, and storage medium for analyzing building occupancy rate data. It can filter normal sampling point data within the fenced area of ​​a building, and then expand the sampling area of ​​the building and its surroundings to increase the number and scope of sampling points. Multi-domain data representing occupancy characteristics from the expanded sampling point data is input into a multi-domain data fusion model to calculate the target building's occupancy rate. The priority of user perception protection is determined based on the building's occupancy rate, achieving automated processing of perception protection, improving work efficiency, and achieving cost reduction and efficiency improvement.

[0006] In a first aspect, embodiments of this application provide a method for analyzing occupancy rate data of a residential building, comprising: acquiring sampling point data from a user terminal, wherein the sampling point data includes communication data and government data; constructing an anomaly detection model for the target residential building using a single-class support vector machine algorithm; inputting the sampling point data and the fence area data of the target residential building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target residential building, thereby obtaining normal sampling point data within the fence area of ​​the target residential building, wherein the target residential building is pre-identified; performing regional expansion processing on the normal sampling point data using a trained coverage feature model to obtain expanded sampling point data; inputting the fence area data of the target residential building and the expanded sampling point data into a trained multi-domain data fusion model for occupancy rate analysis, and outputting the occupancy rate of the target residential building.

[0007] In one possible implementation, the step of constructing an anomaly detection model for the target building using a single-class support vector machine algorithm, and inputting the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building, includes: aggregating the sampling points within the fence area of ​​the target building using the intersection method based on the sampling point data to obtain target sampling point data; determining the main service area and adjacent service areas corresponding to the target building based on the target sampling point data, and determining the voltage level value corresponding to the main service area and the adjacent service areas. The signal strength values ​​corresponding to the adjacent service areas are used to characterize the signal strength of the main service area or the adjacent service area. The main service area and the adjacent service area are grouped according to each group, and a signal strength feature vector is constructed using the signal strength values ​​of the main service area and the adjacent service area in each group. A single-class support vector machine algorithm is used to train the signal strength feature vector model to obtain a trained anomaly detection model. The sampling point data and the fence area data of the target building are input into the anomaly detection model to detect normal sampling points, thereby obtaining normal sampling point data within the fence area of ​​the target building.

[0008] In one possible implementation, the step of using a trained coverage feature model to perform regional expansion processing on the normal sampling point data to obtain expanded sampling point data includes: determining the main service area and corresponding main service level intensity value of the target building based on the normal sampling point data; determining the target number of adjacent service areas adjacent to the main service area and their corresponding adjacent level intensity values; determining the average level intensity value based on the main service level intensity value and the adjacent level intensity values; constructing a coverage feature model database for all buildings based on the main service area, the adjacent service areas, and the average level intensity value; inputting the level intensity value corresponding to the normal sample point data and the coverage feature model database into the coverage feature model, and outputting expanded sampling point data, wherein the expanded sampling point data is the normal sample point data whose level intensity value is within a set level threshold range.

[0009] In one possible implementation, after obtaining the expanded sampling point data, the method further includes: statistically analyzing the feature data within the fenced area of ​​each building using unit time granularity, the feature data including unit time granularity sampling points, number of active users, number of resident users, traffic data, and call volume data; generating a two-dimensional time series set based on the feature data and the set time axis of the unit time granularity; obtaining data labels for each building within a historical time period based on the unique identifier of each building, the data labels indicating whether there are residents in each building; training a binary classification prediction model based on the two-dimensional time series set and the data labels to obtain a trained binary classification judgment model; inputting the feature data of the target building into the binary classification judgment model and outputting the occupancy judgment result of the target building; and, if the occupancy judgment result indicates that there are residents in the target building, performing the step of inputting the fenced area data of the target building and the expanded sampling point data into the trained multi-domain data fusion model for occupancy rate analysis.

[0010] In one possible implementation, the step of inputting the fenced area data of the target building and the expanded sampling point data into a trained multi-domain data fusion model for occupancy rate analysis and outputting the occupancy rate of the target building includes: acquiring the location data and government data of each resident in the expanded sampling point data within the fenced area of ​​the target building, wherein the government data includes electricity data, water data, and gas data; inputting the training set of the location data and the training set of the government data into the multi-domain data fusion training model for training to obtain a trained multi-domain data fusion model; and inputting the test set of the location data and the test set of the government data of each resident into the multi-domain data fusion model to output the occupancy rate of the target building.

[0011] In one possible implementation, the target property is determined by: acquiring a remote sensing image layer within a selected area; rasterizing the remote sensing image layer according to a set size based on the longitude and latitude data in the remote sensing image layer to obtain a target dataset; inputting the target dataset into a pre-trained property recognition model to perform property recognition and outputting the property recognition result; and determining the target property based on the property recognition result.

[0012] In one possible implementation, the step of inputting the target dataset into a pre-trained building recognition model for building recognition and outputting building recognition results includes: obtaining building marker data after the target dataset has been labeled; inputting the training set of the building marker data into a building recognition training model constructed using a target detection algorithm for training to obtain a trained building recognition model; and inputting the test set of the building marker data into the building recognition model to output building recognition results.

[0013] Secondly, embodiments of this application provide a data analysis device for building occupancy rates. The device includes: an acquisition module for acquiring sampling point data from a user terminal, the sampling point data including communication data and government data; an anomaly detection module for constructing an anomaly detection model for a target building using a single-class support vector machine algorithm, and inputting the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building, wherein the target building is pre-identified; an expansion module for performing regional expansion processing on the normal sampling point data using a trained coverage feature model to obtain expanded sampling point data; and an occupancy rate analysis module for inputting the fence area data of the target building and the expanded sampling point data into a trained multi-domain data fusion model for occupancy rate analysis, and outputting the occupancy rate of the target building.

[0014] Thirdly, embodiments of this application provide an analysis device based on building occupancy rates, the device comprising: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the building occupancy rate data analysis method as described in any one of the first aspects.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the building occupancy rate data analysis method as described in any one of the first aspects.

[0016] The occupancy rate data analysis method, apparatus, device, and storage medium of this application embodiment acquire sampling point data from user terminals; input the sampling point data and the fence area data of the target building into an anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby filtering normal sampling point data for the fence area of ​​the building; use a trained coverage feature model to perform regional expansion processing on the normal sampling point data, and increase the number and range of sampling points by expanding the area of ​​the building and its surrounding areas; input multi-domain data that can characterize occupancy characteristics from the fence area data of the target building and the expanded sampling point data into a multi-domain data fusion model for occupancy rate analysis, output the occupancy rate of the target building, and determine the priority of user perception protection based on the occupancy rate, thereby realizing automated processing of perception protection, improving work efficiency, and achieving the effect of cost reduction and efficiency improvement. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for analyzing building occupancy rate data provided in an embodiment of this application; Figure 2 This is a flowchart illustrating a method for determining normal sampling point data, provided in an embodiment of this application. Figure 3 This is a schematic diagram illustrating the effect of an anomaly detection model provided in an embodiment of this application; Figure 4 This is a schematic diagram of a process for determining amplification point data provided in an embodiment of this application; Figure 5 This is a flowchart illustrating a method for identifying occupancy in a residential building, as provided in an embodiment of this application. Figure 6 This is a schematic diagram of the training process of a binary classification prediction model provided in an embodiment of this application; Figure 7 This is a schematic diagram of a process for determining the occupancy rate of a building, provided in an embodiment of this application. Figure 8 This is a block diagram illustrating how to determine the occupancy rate of a building, as provided in an embodiment of this application. Figure 9a This is a flowchart illustrating a method for identifying buildings provided in an embodiment of this application; Figure 9b This is a schematic diagram of a remote sensing image layer provided in an embodiment of this application; Figure 9c This is a schematic diagram of the rasterized structure of a remote sensing image provided in an embodiment of this application; Figure 10a This is a flowchart illustrating another method for identifying buildings provided in an embodiment of this application; Figure 10b This is a schematic diagram of remote sensing image labeling provided in an embodiment of this application; Figure 10c This is a rendering of a building identification system provided in an embodiment of this application; Figure 11 This is a flowchart illustrating another method for analyzing building occupancy rate data provided in an embodiment of this application; Figure 12 This is a schematic diagram of the structure of a building occupancy rate data analysis device provided in an embodiment of this application; Figure 13 This is a schematic diagram of the structure of a building occupancy rate data analysis device provided in an embodiment of this application. Detailed Implementation

[0019] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0020] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0021] Communication sensing assurance refers to a technical system that achieves real-time monitoring and optimization of communication service quality, network status, and environmental information through the deep integration of communication networks and sensing technologies, thereby ensuring the stable and reliable operation of communication systems.

[0022] Currently, the system identifies resident users in a target building based on the historical connection count of each base station in the target building's base station list. When a resident user is detected entering the target building, the system acquires the signal strength data of the base station connected to that user. The network quality of the target building is then assessed based on this signal strength data. The effectiveness of communication awareness assurance is judged by the network quality. However, the existing solution is relatively slow and cannot provide timely awareness assurance. Furthermore, the current solution relies heavily on manual operation during information analysis, which is time-consuming and labor-intensive. This leads to insufficient accuracy in analyzing occupancy scale.

[0023] To address the problems of existing technologies, this application provides a method, apparatus, device, and storage medium for analyzing building occupancy rate data. By comparing sampling point data obtained from user terminals within the identified fenced area of ​​a building, sampling point data outside the fenced area is removed. For the sampling point data within the fenced area, a coverage feature model is used to expand the sampling area, resulting in expanded sampling point data. The occupancy rate is calculated using a multi-domain data fusion model based on the multi-domain data contained in the expanded sampling point data. The occupancy rate reflects the user's communication awareness and security capabilities, improving the accuracy of occupancy rate analysis. The method for analyzing building occupancy rate data provided in this application is described below.

[0024] Figure 1 This illustration shows a flowchart of a method for analyzing building occupancy rate data according to an embodiment of this application. Figure 1 As shown, the specific steps of the data analysis method for building occupancy rate include: S101~S104.

[0025] S101. Obtain sampling point data from the user terminal. The sampling point data includes communication data and government data.

[0026] The sampling point data mentioned here can be understood as including network data and daily consumption data collected from user terminal devices. The communication data mentioned here can be understood as Minimization Drive Test (MDT) data. The government data mentioned here can be understood as water, electricity, gas, etc., bills for daily life.

[0027] Furthermore, network data and living expense data of end users are obtained. By using network usage data and living expense data, user sampling point data is determined, providing reference data for the next step of determining whether a user is a resident of a designated building.

[0028] S102. Construct an anomaly detection model for the target building using a single-class support vector machine algorithm, and input the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building. The target building is pre-identified and determined.

[0029] The single-class support vector machine algorithm mentioned here can be understood as the One-Class-SVM algorithm. One-Class-SVM is an unsupervised learning algorithm specifically designed for anomaly detection and novelty detection. Unlike traditional One-Class-SVM, it is trained using only data from a single class and identifies anomalies by constructing decision boundaries in a high-dimensional space. The anomaly detection model can be understood as detecting sampling data outside the designated fenced area of ​​a building complex. The fenced area can be understood as the region defined by the longitude and latitude values ​​of the building complex. The target building can be understood as the building identified or located through other means.

[0030] Furthermore, after obtaining the user sampling point data, to improve the calculation accuracy of the occupancy rate of the selected target building, it is necessary to select the sampling point data within the fenced area of ​​the target building for judgment. The fenced area is selected using the longitude and latitude data of the target building. An anomaly detection model, trained to determine whether each sampling point belongs to the fenced area of ​​the target building, removes sampling points outside the fenced area, and obtains the sampling point data within the fenced area (including sampling point data on the boundary). The obtained sampling point data within the fenced area is considered normal sampling point data. This provides more accurate sampling point data for the next step of analyzing the occupancy rate of the target building, reduces the computational load of all sampling point data, and improves computational processing efficiency.

[0031] S103. Use the trained coverage feature model to perform regional expansion processing on the normal sampling point data to obtain expanded sampling point data.

[0032] The coverage feature model mentioned here can be understood as a feature comparison model that increases the range of sampling point data through feature comparison.

[0033] The expanded sampling point data mentioned here can be understood as an increase in the number of sampling points compared to normal sampling point data, while the distribution area of ​​the sampling point data also increases.

[0034] Furthermore, to improve the accuracy of occupancy rate detection for target properties, it is necessary to increase the number and scope of sampling points based on the normal sampling point data. This is done by inputting the selected normal sampling point data into a trained coverage feature model, and then matching the normal sampling point data with the feature data set in the coverage feature model to increase the number of sampling points that meet the feature data requirements, thereby expanding the sampling point range and obtaining expanded sampling point data. By expanding the range and number of sampling points, more potentially occupied sampling point data is added for the next step of analyzing the occupancy rate of the target property, thus improving the detection accuracy.

[0035] S104. Input the fence area data and expansion point data of the target building into the trained multi-domain data fusion model to analyze the occupancy rate and output the occupancy rate of the target building.

[0036] The multi-domain data fusion model mentioned here can be understood as a process of matching based on the feature data contained in the sampling point data. Sampling points that meet the matching conditions are identified as occupant users, thereby realizing occupancy rate analysis.

[0037] Further, after obtaining the expanded sampling point data, multi-domain feature data representing user occupancy status is acquired from this data. This multi-domain feature data is then input into a multi-domain data fusion model to calculate eligible sample points within the fenced area of ​​the target building. The finally detected sample points that meet the criteria are output as occupant users, thus obtaining the occupancy rate for the target building. By using multi-domain data to determine permanent residents, the occupancy status is detected more precisely, providing accurate calculations for determining the building's occupancy rate. Based on the building's occupancy rate, the priority of perception and protection is determined, achieving fully automated processing of the perception and protection process, significantly improving work efficiency and achieving the technical effect of cost reduction and efficiency improvement.

[0038] This application provides a method for analyzing occupancy rate data of a residential building. The method involves acquiring sampling point data from user terminals; inputting the sampling point data and the fence area data of the target building into an anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thus filtering normal sampling point data for the fence area; using a trained coverage feature model to perform regional expansion processing on the normal sampling point data, increasing the number and range of sampling points by expanding the area of ​​the building and its surroundings; inputting multi-domain data representing occupancy characteristics from the fence area data and expanded sampling point data of the target building into a multi-domain data fusion model for occupancy rate analysis, outputting the occupancy rate of the target building; and determining the priority of user perception protection based on the occupancy rate, thereby achieving automated processing of perception protection, improving work efficiency, and achieving cost reduction and efficiency improvement.

[0039] In one possible example scenario, the implementation process of step S102, which determines normal sampling point data based on sampling point data, is as follows: Figure 2 This is a schematic diagram of a process for determining normal sampling point data provided in an embodiment of this application. According to... Figure 2 The steps for determining normal sampling point data, as shown in the provided diagram, include: S201~S205.

[0040] S201. Based on the sampling point data, the intersection method is used to aggregate the sampling points within the fenced area of ​​the target building to obtain the target sampling point data.

[0041] To improve the accuracy of occupancy rate calculation, it is necessary to filter the sampling point data within the fenced area of ​​the building and remove sampling points that do not belong to the fenced area of ​​the building as abnormal sampling point data, thereby reducing the impact of abnormal sampling point data when calculating the occupancy rate.

[0042] The intersection method mentioned here can be understood as drawing a ray from a point through an irregular polygon, eliminating vertices, boundary points, etc. If the number of intersection points is odd, the polygon is inside; if it is even, the polygon is outside.

[0043] Furthermore, for the collected sampling point data, the minimum road test MDT data contained in the sampling point data is combined with the latitude and longitude data of the target building's location to form a fenced area. The intersection method is used to aggregate the sampling points of the target building within the fenced area as target sampling points, thus obtaining target sampling point data.

[0044] S202. Based on the target sampling point data, determine the main service area and adjacent service area corresponding to the target building, and determine the level strength value corresponding to the main service area and the level strength value corresponding to the adjacent service area. The level strength value is used to characterize the signal strength of the main service area or the adjacent service area.

[0045] S203. Group the service areas according to each primary service area and adjacent service areas, and construct a level intensity feature vector using the level intensity values ​​of the primary service area and adjacent service areas in each group.

[0046] The primary service area, as referred to here, can be understood as the signal range covered by the primary base station set for the target building. The adjacent service area can be understood as the additional signal range covered by adjacent base stations relative to the primary base station. The voltage level value, as referred to here, can be understood as the different voltage level values ​​received by the user terminal representing the target sampling point from different base stations. In this application, the number and names of the base stations connected to the user terminal, as well as their voltage level values, are all pre-set.

[0047] Furthermore, based on the selected target sampling points, the main service area and adjacent service areas corresponding to the target building are determined. The voltage level value corresponding to the main service area is determined based on the determined main service area, and the voltage level value corresponding to the adjacent service areas is determined based on the adjacent service areas.

[0048] In one possible scenario, based on the selected target sampling point data within the target building, the main service area and the top 5 adjacent service areas selected by sampling points, along with their corresponding voltage levels, are organized. These are then grouped by the main service area and the top 5 adjacent service areas. Groups with fewer sampling points are directly discarded, while groups with more sampling points are organized into structured vector feature data (vectors composed of the voltage levels of the main service cell and neighboring cells). This vector feature data provides matching data for the next step of detecting abnormal sampling point data.

[0049] S204. Use a single-class support vector machine algorithm to train the model of the level intensity feature vector to obtain a trained anomaly detection model.

[0050] S205. Input the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points and obtain the normal sampling point data within the fence area of ​​the target building.

[0051] The One-Class SVM algorithm mentioned here considers data similar to normal data (i.e., grouped together by the hyperplane) as normal data, and data that differs significantly from normal data as outliers. Based on this idea, for the sample data of each group, One-Class SVM is used for detection, and appropriate boundaries are adjusted. Outlier sampling points with large discrepancies are removed, thus obtaining normal sampling data.

[0052] For example, in one possible scenario, Figure 3 This is a schematic diagram illustrating the effect of an anomaly detection model provided in an embodiment of this application. According to... Figure 3 The provided diagram shows that the anomaly detection model built using the single-class support vector machine algorithm is equivalent to... Figure 3 The circular area is defined as follows: sampling points within the area are considered normal sampling points, while sampling points outside the area are considered abnormal sampling points. This process yields the filtered normal sampling point data.

[0053] The occupancy rate data analysis method provided in this application, for the screening process of normal sampling points, firstly aggregates sampling points within the fenced area that conforms to the target building based on the minimized road test data in the sampling point data using the intersection method, then constructs an anomaly detection model for the main service area and adjacent service areas through user sampling points of the target building, and finally removes abnormal sampling points in the target building according to the anomaly detection model to obtain normal sampling point data.

[0054] In one possible example scenario, the implementation process of step S103, which involves obtaining amplified sample point data from normal sample point data, is as follows: Figure 4 This is a schematic diagram of a process for determining amplification point data provided in an embodiment of this application. According to... Figure 4 The steps for determining the data of the expansion points, as shown in the provided diagram, include: S401~S405.

[0055] S401. Determine the main service area and corresponding main service level intensity value of the target building based on normal sampling point data.

[0056] S402. Determine the target number of adjacent service areas adjacent to the main service area and the corresponding adjacent level strength values.

[0057] S403. Determine the average level strength value based on the main service level strength value and adjacent level strength values.

[0058] S404. Construct a coverage feature model database for all buildings based on the main service area, adjacent service areas, and average power level values.

[0059] The signal strength is used to characterize the strength of the coverage signal corresponding to the base station in the main service area and adjacent service areas.

[0060] The average voltage level is used as the criterion for evaluating whether the voltage level values ​​in the sampled data conform to the coverage feature model database. The average voltage level is then used as the threshold voltage. The target quantity here can be a set quantity or a random quantity.

[0061] Furthermore, after obtaining normal sampling point data, the main base station set in the fenced area of ​​the target building is determined as the main service area, and the signal coverage range corresponding to the main service area is determined by combining the sampling point data. At the same time, for each main service area, five neighboring cells are locked (sorted by sampling points) in the corresponding adjacent service areas. The average level strength of different service areas and the corresponding five adjacent areas is calculated and stored in the form of a dictionary. Then, the building coverage feature model library is constructed by using the obtained location and name of the main service area and adjacent service areas and the corresponding level strength value information.

[0062] In one possible example scenario, a group is formed by the main service area (i.e., the main service cell) and five adjacent service areas within different building complexes. The voltage level value for each area is set. The coverage feature model database shown in Table 1 is obtained.

[0063] Table 1

[0064] S405. Input the level intensity value corresponding to the normal sample point data and the coverage feature model database into the coverage feature model, and output the expanded sample point data. The expanded sample point data is the normal sample point data whose level intensity value is within the set level threshold range.

[0065] By inputting the signal strength values ​​from the sampling point data into the coverage feature model, the signal strength values ​​are compared with those of each main service area and five adjacent service areas contained in the coverage feature model database. If the difference is lower than the average signal strength value, the user terminal receiving the signal is considered to be below the signal strength of the building's coverage, and this sampling point is removed. When the signal strength value of a sampling point is detected to be higher than the average signal strength value, the user terminal's service signal strength is considered to be within the range provided by the building, and this sampling point is designated as an expansion point. The expansion point data is then output through the coverage feature model.

[0066] In one possible scenario, based on the data provided in Table 1, to determine whether a sampling point is within a building complex, first, it is determined whether the main service cell of the sampling point is in the list of the building's coverage feature model database. If not, it does not belong to that building complex. If it is in list 1 of the coverage feature model database and the difference between the voltage level of the sampling point data and the voltage level in the coverage feature model database is within 5 dBm (this value can be set differently depending on the size of the building complex), then the neighboring service areas of the sampling point are determined. If four or more of the five neighboring service areas in the coverage feature model database appear in the neighboring service areas provided by the sampling point and the difference between the voltage level and the voltage level in the coverage feature model database is within 5 dBm, then the sampling point is determined to belong to that building complex; otherwise, it does not belong to that building complex, and the sampling point is used as the expanded sampling point data.

[0067] The occupancy rate data analysis method for residential buildings provided in this application is based on normal sampling point data combined with the coverage feature model database of the residential buildings. If the main service area and the selected corresponding adjacent area are consistent and the level intensity value fluctuates within a certain range, then the sampling point is determined to belong to the user sampling point of the residential building. In this way, the sampling of user sampling points of the residential building is expanded, making the data source for analysis more comprehensive.

[0068] In one possible example scenario, after obtaining the expanded sample point data, before processing the expanded sample point data using a multi-domain data fusion model, there is also a step of identifying the occupancy of the building. Figure 5 This is a flowchart illustrating a method for identifying occupancy in a residential building, as provided in an embodiment of this application. According to... Figure 5 The provided diagram illustrates the steps for identifying occupancy in a residential building, specifically including: S501~S505.

[0069] S501. Utilize unit time granularity to statistically analyze the characteristic data within the fenced area of ​​each building. The characteristic data includes unit time granularity sampling points, number of active users, number of resident users, traffic data, and call volume data.

[0070] S502. Generate a two-dimensional time series set based on the feature data and the set time axis with unit time granularity.

[0071] The unit time granularity mentioned here can be understood as the set time interval as the detection cycle. For example, granularity of hours, days, months, years, etc.

[0072] Based on sampling point data collected from user terminals and combined with the latitude and longitude data of the building, the location of the building is determined. Within the fenced area of ​​the target building, MDT sampling point data is statistically analyzed on an hourly basis. Additionally, based on the fenced area of ​​the target building, the number of active users within the fenced area is statistically analyzed on an hourly basis (deduplicated user count within the fenced area). The number of resident users within the fenced area is also statistically analyzed on an hourly basis (the number of consecutive days of user activity within the fenced area is considered resident users). User traffic within the fenced area is also statistically analyzed on an hourly basis. Finally, based on the fenced area data, user call volume within the fenced area is statistically analyzed on an hourly basis. This process results in the collection of a feature dataset.

[0073] Based on the obtained feature dataset, a two-dimensional time series dataset is generated by combining hourly time parameters. In one possible scenario, the structure of the feature dataset for a single property is shown in Table 2. Table 2

[0074] S503. Obtain the data label for each building within the historical time period based on the unique identifier of each building. The data label indicates whether there are residents in each building. Based on the two-dimensional time series set and the data label, train the binary classification prediction model to obtain the trained binary classification judgment model.

[0075] The unique identifier for a property here refers to its property ID number or other identifiers that can distinguish it. The data labels can be understood as assigning positive or negative labels (1 for occupancy, 0 for no occupancy) to known samples based on historical occupancy timeframes. This results in the feature dataset table 3 with data labels. Table 3

[0076] Note: The feature dataset here is a two-dimensional data frame (the table above only presents one-dimensional data features and does not show the two-dimensional time series). The data labels are one-dimensional numerical values. The feature data and data labels are stored in two separate files. The model is trained and predicted using a unique building ID identifier to link them.

[0077] Based on the feature data and data labels provided by the sampling point data, a binary classification prediction model is trained to establish a binary classification prediction model based on a convolutional neural network. After training, a binary classification judgment model is obtained. This model can determine whether a building has residents. It only determines the presence of residents; no special collection of specific user data is required, simplifying the recognition process and improving efficiency.

[0078] For example, in one possible scenario, Figure 6 This is a schematic diagram illustrating the training process of a binary classification prediction model provided in an embodiment of this application. According to... Figure 6 The provided diagram illustrates how data on the fenced area of ​​the target building and collected feature data are input into a binary classification prediction model containing the post-feature dataset. The model is trained using convolution and pooling operations on the data with labels and features, and the recognition result is used as the model output. This results in a trained binary classification judgment model (i.e.,...). Figure 6 (A dense crowd algorithm model).

[0079] S504. Input the feature data of the target building into the binary classification judgment model and output the occupancy judgment result of the target building.

[0080] S505. If the occupancy determination result indicates that there are occupants in the target building, perform the step of inputting the fenced area data and expanded sampling point data of the target building into the trained multi-domain data fusion model for occupancy rate analysis.

[0081] By inputting the feature data into a binary classification model, the determination of whether a target building is occupied can be obtained. Specifically, two-dimensional feature information (hourly granular MDT sampling points, active users, resident users, traffic, call volume, and related indicators are organized into structured two-dimensional data) is input into the binary classification model. The model outputs the probability of whether someone has moved in (the prediction result can also be formatted into 0 / 1 based on a threshold; for example, a probability greater than 0.5 indicates occupancy). For example, the output results are shown in Figure 4. Table 4

[0082] To further verify the effectiveness of the binary classification model, we selected properties in a specific market grid area for testing. Over 300 properties were randomly selected for testing, and the model's prediction results are shown in Table 5. Table 5

[0083] Based on the data provided in Table 5, the overall accuracy rate reached (212+123) / 375=89.33%, which shows that the model prediction effect met the set requirements and the accuracy was improved, indicating that the accuracy of the binary classification judgment model constructed in this application met the requirements.

[0084] Based on the occupancy rate data analysis method provided in this application, a binary classification prediction model is constructed using a two-dimensional time series set and data labels on the basis of selected expansion point data. After training the model, a binary classification judgment model is obtained, which can determine whether there are residents based on the feature data of the building user end, simplifying the process of identifying user occupancy.

[0085] In one possible example scenario, step S104 involves calculating the occupancy rate of the target building based on the expanded sampling point data. Figure 7 This is a schematic diagram illustrating a process for determining the occupancy rate of a residential building, provided in an embodiment of this application. According to... Figure 7 The provided diagram shows the specific steps involved in determining the occupancy rate of the property, including steps S701 to S703.

[0086] S701. Obtain the location data and government data of each resident in the expanded sampling point data within the fenced area of ​​the target building. The government data includes electricity data, water data and gas data.

[0087] The government data mentioned here can be understood as data used to characterize users' residential attributes, and includes not only electricity data, water data, and gas data.

[0088] Furthermore, assuming there are residents in the target building, it is necessary to statistically analyze occupancy data to determine the occupancy rate. Only after assessing whether residents meet the criteria for permanent residency can they be included in the target building's resident statistics. This involves first acquiring user location data and government data, then extracting the building's location distribution and related resident information, including daily-level time-based data collection of each resident's electricity, water, and gas consumption.

[0089] S702. Input the training sets of location data and government data into the multi-domain data fusion training model for training to obtain the trained multi-domain data fusion model.

[0090] S703. Input the test set of location data for each resident user and the test set of government data into the multi-domain data fusion model, and output the occupancy rate of the target building.

[0091] Government data is used as multi-domain data. Training sets for both location data and government data are input into a multi-domain data fusion training model for training, resulting in a trained model. Based on the daily consumption habits of residents (water, electricity, gas, etc.) and combined with multi-day data collection, a multi-domain data fusion model for resident occupancy identification is established to predict whether each resident is occupant normally (the model excludes scenarios involving water and electricity usage such as interior decoration and installation). Based on the multi-domain data fusion model, the prediction of individual resident occupancy is used to count the number of normally occupant residents. This count, combined with the total number of residents in the building, is used to calculate the occupancy rate. The occupancy rate determines the priority of security measures, guiding frontline staff to rationally allocate security work.

[0092] For example, in one possible scenario, Figure 8 This is a block diagram illustrating how to determine the occupancy rate of a residential building, as provided in an embodiment of this application. According to... Figure 8 The provided diagram first performs anomaly detection on the sampling point data within the fenced area of ​​the selected building complex. By establishing an anomaly detection model, sampling points that do not belong to the fenced area of ​​the building complex are removed, resulting in normal sampling point data. The normal sampling point data undergoes regional expansion processing to obtain expanded sampling point data, increasing the recognition range. Then, a binary classification judgment model is used to determine whether there are resident users in the building complex. If so, the government data of the sampling point data within the fenced area of ​​the building complex is further obtained. By processing the government data from multiple domains using a multi-domain data fusion model, the occupancy rate of the resident users in the building complex is obtained, further improving the user's communication perception guarantee effect.

[0093] The occupancy rate data analysis method for residential buildings provided in this application involves detecting abnormal sampling points within the fenced area of ​​the residential building to obtain normal sampling point data within the fenced area; then, expanding the sampling data through expansion processing; and finally, using a binary classification judgment model to determine whether there are occupants in the residential building. If there are occupants, the government data of the expanded sampling point data within the fenced area of ​​the current residential building is subjected to multi-domain data fusion processing to output the occupancy rate data of the residential building, thereby improving user communication perception and security.

[0094] Based on the occupancy rate of the property, the process of identifying the property is as follows: Figure 9a This is a flowchart illustrating a method for identifying building types according to an embodiment of this application. Figure 9a The provided diagram illustrates the specific steps for identifying the property, including S901~S904.

[0095] S901. Obtain the remote sensing image layer within the selected area.

[0096] Obtain remote sensing image layers from map providers by city or region, or by selecting a region according to requirements. The layers contain relevant information such as region boundaries, high-resolution images, and latitude and longitude.

[0097] In one possible example scenario, Figure 9b This is a schematic diagram of a remote sensing image layer provided in an embodiment of this application. Obtaining a remote sensing image layer facilitates the identification of actual building location information, including latitude and longitude information and marker information.

[0098] S902. Based on the longitude and latitude data in the remote sensing image layer, the remote sensing image layer is rasterized according to the set size to obtain the target dataset.

[0099] To quickly identify buildings in a remote sensing image layer, the entire image needs to be segmented into multiple smaller images according to a set size, resulting in a target dataset containing multiple images.

[0100] In one possible example scenario, Figure 9c This is a schematic diagram of the rasterized structure of a remote sensing image provided in an embodiment of this application. According to... Figure 9c The provided illustration shows the image results obtained after rasterizing the selected remote sensing image layer. Rasterization and cropping are performed along both longitude and latitude directions at appropriate fixed sizes to generate images of fixed sizes, while simultaneously associating them with the latitude and longitude information of the corresponding layer. Building recognition methods based on remote sensing imagery have applications in other fields. In the construction field, processed remote sensing images are generally divided into four stages: foundation stage, construction stage, topping-out stage, and completion stage. Image classification models are established to predict the construction progress of buildings, primarily using image classification techniques. However, this application mainly employs target detection technology to identify new buildings within remote sensing image layers, differing from conventional foundation-based classification methods.

[0101] S903. Input the target dataset into the pre-trained building recognition model to identify the building and output the building recognition result.

[0102] S904. Determine the target building based on the building identification results.

[0103] The building identification model mentioned here can be understood as a model trained using the YOLO object detection algorithm. This model directly outputs the latitude and longitude data of the building, thus selecting it as the target building and providing a basis for calculating its occupancy rate in the next step.

[0104] Regarding the process of identifying the property in step S903, Figure 10a This is a flowchart illustrating another method for identifying building types provided in an embodiment of this application. According to... Figure 10aThe provided diagram illustrates the steps for identifying buildings using a building identification model, including: S1001~S1003.

[0105] S1001. Obtain the building label data after the target dataset has been labeled.

[0106] The target dataset, generated from rasterized cropping of remote sensing images, is labeled using the labelImg annotation tool to identify planned or under-construction new buildings in the images. The labeled files are then output as files in a specified format (e.g., XML files containing the location information of the buildings).

[0107] In one possible example scenario, Figure 10b This is a schematic diagram of remote sensing image labeling provided in an embodiment of this application. According to... Figure 10b The provided image allows users to label buildings within the selected area using the labelImg tool, such as by name (e.g., school, hospital, residential building). This generates building marker data with annotation information, providing a basis for the next step of identifying buildings.

[0108] S1002. Input the training set of building marker data into the building recognition training model constructed using the object detection algorithm for training, and obtain the trained building recognition model.

[0109] S1003. Input the test set of building marker data into the building recognition model and output the building recognition result.

[0110] For example, given an image, the image is first divided into an S*S grid. For each grid, B bounding boxes are predicted (including the confidence score of each bounding box as a target and the probability of each bounding box region across multiple categories). Based on the previous step, S*S*B target windows can be predicted. Then, target windows with low probability are removed according to a threshold. Finally, redundant windows are removed to obtain the recognition result. This process yields the trained building recognition model. The test set of building marker data is input into the building recognition model, and the output is the building recognition result containing location information and confidence scores.

[0111] In one possible example scenario, Figure 10c This is a rendering of a building identification system provided in an embodiment of this application. The system identifies building marker data using a building identification model, and then assigns the latitude and longitude coordinates (building0.92) to the identified building markers.

[0112] Optionally, confidence assessment and latitude / longitude conversion are performed on the output building identification results. For each identification result, a rectangular box is presented in the image, along with its category and confidence level. The confidence level represents the probability that the rectangular box is accurately identified. A threshold can be determined based on the output results; sampling points with confidence levels below the threshold are considered unreliable. The coordinates carried in the identification results represent the pixel positions in the image. In the previous remote sensing image data processing stage, the rasterized and cropped images correspond to latitude and longitude information. Latitude and longitude conversion is performed by combining the image's latitude and longitude data with the rectangular box pixel positions to output the building's latitude and longitude information (i.e., the latitude and longitude coordinates of the four vertices of the rectangular box, not shown in the figure).

[0113] In one possible example scenario, Figure 11 This is a flowchart illustrating another method for analyzing building occupancy rate data provided in an embodiment of this application. According to... Figure 11 The provided diagram first preprocesses the image using a remote sensing image layer to remove noise that affects image quality. Then, it rasterizes the remote sensing image layer according to latitude and longitude data, resulting in multiple regions of equal size. Within each region, a building identification model is used to identify buildings under construction and those already occupied. Sampling points are then selected within the fenced areas of the identified buildings to obtain expanded sampling point data. Feature data representing occupants is then extracted from these expanded sampling point data, and a binary classification model for building occupancy identification is used to identify buildings with occupants. For these buildings, a multi-domain data fusion model is used to calculate the occupancy rate, yielding communication awareness assurance results. Based on the occupancy rate, the priority of awareness assurance is determined, achieving fully automated processing of the entire awareness assurance process, significantly improving work efficiency and achieving cost reduction and efficiency improvement.

[0114] Figure 12 This is a schematic diagram of the structure of a building occupancy rate data analysis device provided in an embodiment of this application. According to... Figure 12 The provided diagram shows that the building occupancy rate data analysis device 100 specifically includes: 121~124.

[0115] The acquisition module 121 is used to acquire sampling point data from the user terminal, including communication data and government data. The anomaly detection module 122 is used to construct an anomaly detection model of the target building using a single-class support vector machine algorithm, and to input the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building. The target building is pre-identified and determined. The expansion module 123 is used to perform regional expansion processing on the normal sampling point data using the trained coverage feature model to obtain expanded sampling point data. The occupancy rate analysis module 124 is used to input the fence area data and expansion point data of the target building into the trained multi-domain data fusion model to perform occupancy rate analysis and output the occupancy rate of the target building.

[0116] The building occupancy rate data analysis device 100 provided in this embodiment can be as follows: Figure 12 The occupancy rate data analysis device 100 shown can achieve... Figures 1-11 For a detailed description of the technical effectiveness of the occupancy rate data analysis method for the shown properties, please refer to [link / reference]. Figures 1-11 The corresponding explanation is concise and will not be elaborated upon here.

[0117] Figure 13 This is a schematic diagram of the structure of a building occupancy rate data analysis device provided in an embodiment of this application. The building occupancy rate data analysis device may include a processor 1301 and a memory 1302 storing computer program instructions.

[0118] Specifically, the processor 1301 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0119] Memory 1302 may include mass storage for data or instructions. For example, and not limitingly, memory 1302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 1302 may include removable or non-removable (or fixed) media. Where appropriate, memory 1302 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 1302 is non-volatile solid-state memory.

[0120] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0121] The processor 1301 reads and executes computer program instructions stored in the memory 1302 to implement any of the building occupancy rate data analysis methods in the above embodiments.

[0122] In one example, the building occupancy rate data analysis device may also include a communication interface 1303 and a bus 1310. For example, Figure 13 As shown, the processor 1301, memory 1302, and communication interface 1303 are connected through bus 1310 and complete communication with each other.

[0123] The communication interface 1303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0124] Bus 1310 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 1310 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0125] Furthermore, in conjunction with the property occupancy rate data analysis method in the above embodiments, this application embodiment can provide a computer storage medium for implementation. This computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the property occupancy rate data analysis methods in the above embodiments.

[0126] This application also provides a computer program product, including a computer program, which, when executed, implements any of the building occupancy rate data analysis methods described in the above embodiments.

[0127] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0128] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0129] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0130] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0131] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for analyzing occupancy rate data of a residential building, characterized in that, include: Acquire sampling point data from user terminals, the sampling point data including communication data and government data; An anomaly detection model for the target building is constructed using a single-class support vector machine algorithm. The sampling point data and the fence area data of the target building are input into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building. The target building is pre-identified. The normal sampling point data is subjected to regional expansion processing using the trained coverage feature model to obtain expanded sampling point data; The fence area data of the target building and the expansion point data are input into the trained multi-domain data fusion model to analyze the occupancy rate and output the occupancy rate of the target building.

2. The method according to claim 1, characterized in that, The method involves constructing an anomaly detection model for the target building using a single-class support vector machine algorithm, and inputting the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building, including: Based on the sampling point data, the sampling points within the fenced area of ​​the target building are aggregated using the intersection method to obtain the target sampling point data; Based on the target sampling point data, the main service area and adjacent service area corresponding to the target building are determined, and the signal strength value corresponding to the main service area and the signal strength value corresponding to the adjacent service area are determined. The signal strength value is used to characterize the signal strength of the main service area or the adjacent service area. Grouping according to each primary service area and each adjacent service area, and constructing a level intensity feature vector using the level intensity values ​​of the primary service area and the adjacent service area in each group; The level intensity feature vector model is trained using a single-class support vector machine algorithm to obtain a trained anomaly detection model. The sampling point data and the fence area data of the target building are input into the anomaly detection model to detect normal sampling points, thereby obtaining normal sampling point data within the fence area of ​​the target building.

3. The method according to claim 2, characterized in that, The process of using a trained coverage feature model to perform regional expansion processing on the normal sampling point data to obtain expanded sampling point data includes: Based on the normal sampling point data, the main service area and corresponding main service level intensity value of the target building are determined; Determine the target number of adjacent service areas adjacent to the main service area and the corresponding adjacent voltage level values; The average service level strength value is determined based on the primary service level strength value and the adjacent service level strength values. A database of coverage feature models for all buildings is constructed based on the main service area, the adjacent service areas, and the average level intensity value. The level intensity value corresponding to the normal sample point data and the coverage feature model database are input into the coverage feature model, and the expanded sample point data is output. The expanded sample point data is the normal sample point data whose level intensity value is within the set level threshold range.

4. The method according to claim 1, characterized in that, After obtaining the amplification point data, the method further includes: The system uses unit-time granularity to collect characteristic data within the fenced area of ​​each building complex. The characteristic data includes unit-time granularity sampling points, number of active users, number of resident users, traffic data, and call volume data. A two-dimensional time series set is generated based on the feature data and the set time axis with the unit time granularity; Data tags for each building are obtained within a historical time period based on the unique identifier of each building. The data tags indicate whether there are residents in each building. Based on the two-dimensional time series set and the data tags, a binary classification prediction model is trained to obtain a trained binary classification judgment model. The feature data of the target building is input into the binary classification judgment model, and the occupancy judgment result of the target building is output. If the occupancy determination result indicates that there are occupants in the target building, the step of inputting the fence area data of the target building and the expansion point data into the trained multi-domain data fusion model for occupancy rate analysis is performed.

5. The method according to claim 1, characterized in that, The step of inputting the fence area data of the target building and the expanded sampling point data into a trained multi-domain data fusion model for occupancy rate analysis, and outputting the occupancy rate of the target building, includes: Obtain the location data and government data of each resident user in the expanded sampling point data within the fenced area of ​​the target building, wherein the government data includes electricity data, water data and gas data; The training sets of the location data and the government data are input into the multi-domain data fusion training model for training, and a trained multi-domain data fusion model is obtained. The test set of location data and the test set of government data for each resident user are input into the multi-domain data fusion model, and the occupancy rate of the target building is output.

6. The method according to claim 1, characterized in that, The target property was determined in the following way: Obtain the remote sensing image layer within the selected area; Based on the longitude and latitude data in the remote sensing image layer, the remote sensing image layer is rasterized according to a set size to obtain the target dataset; The target dataset is input into a pre-trained building identification model to identify buildings and output the building identification results. The target property is determined based on the property identification results.

7. The method according to claim 6, characterized in that, The step of inputting the target dataset into a pre-trained building identification model for building identification and outputting building identification results includes: Obtain the labeled building data of the target dataset; The training set of the building marker data is input into the building identification training model constructed using the object detection algorithm for training, and a trained building identification model is obtained. The test set of the building marker data is input into the building identification model, and the building identification result is output.

8. A device for analyzing occupancy rate data of a residential building, characterized in that, The device includes: The acquisition module is used to acquire sampling point data from the user terminal, the sampling point data including communication data and government data; An anomaly detection module is used to construct an anomaly detection model for the target building using a single-class support vector machine algorithm, and to input the sampling point data and the fence area data of the target building into the anomaly detection model to detect normal sampling points within the fence area of ​​the target building, thereby obtaining normal sampling point data within the fence area of ​​the target building. The target building is pre-identified and determined. The expansion module is used to perform regional expansion processing on the normal sampling point data using a trained coverage feature model to obtain expanded sampling point data. The occupancy rate analysis module is used to input the fence area data of the target building and the expansion point data into a trained multi-domain data fusion model to perform occupancy rate analysis and output the occupancy rate of the target building.

9. An analysis device based on building occupancy rate, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the building occupancy rate data analysis method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the building occupancy rate data analysis method as described in any one of claims 1 to 7.

Images