Key person position determination method, device and equipment

Abstract

The application relates to the technical field of personnel position determination, in particular to a key personnel position determination method, device and equipment, wherein the accuracy of historical personnel position determination, a historical average time attenuation factor and an average spatial correlation degree are used to dynamically and adaptively determine the weight corresponding to the perception data type. Based on the determined weight corresponding to the perception data type, the time attenuation factor can improve the time sensitivity of the personnel position determination, and the address confidence degree is fused to finally accurately determine the comprehensive score of the address, and the key personnel position is determined based on the comprehensive score of the address, so that the working efficiency can be improved, the personnel cost can be reduced, and the response speed can be improved.

Description

Technical Field

[0001] This application relates to the technical field of personnel location determination, and in particular to a method, apparatus and equipment for determining the location of key personnel. Background Technology

[0002] In the management of key personnel in the field of public security, the current determination of the location of key personnel still mainly relies on passive management methods such as manual visits and offline registration. This is not only inefficient and costly, but also has obvious response delays. Summary of the Invention

[0003] This application provides a method, apparatus, and device for determining the location of key personnel. The method dynamically and adaptively determines the weights corresponding to different types of sensed data based on the historical accuracy of personnel location determination, the historical average time decay factor, and the average spatial correlation. Based on the determined weights corresponding to the sensed data types, fusing the time decay factor can improve the time sensitivity of personnel location determination, and fusing address confidence can accurately determine the comprehensive address score. The location of key personnel is then determined based on the comprehensive address score. This not only improves work efficiency and reduces personnel costs but also increases response speed.

[0004] In a first aspect, embodiments of this application provide a method for determining the location of key personnel, including: Acquire dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types. Obtain the weight value of the perception data type; wherein the step of determining the weight value of the perception data type includes: when a preset condition is met, determining the effective sample size of historical perception data of different perception data types; if the effective sample size of historical perception data of the perception data type is lower than the corresponding preset number, then setting the weight value of the perception data type to a preset static weight value; if the effective sample size of historical perception data of the perception data type is not lower than the corresponding preset number, then setting the weight value of the perception data type based on the historical accuracy of determining personnel location, historical average time decay factor, and average spatial correlation of the perception data type; normalizing the weight values ​​of all perception data types so that the sum of all normalized weight values ​​is equal to the total number of types of all perception data types; smoothly transitioning the normalized weight value with the previously obtained normalized weight value to update the normalized weight value; updating the weight value of the perception data type using the updated normalized weight value; the preset condition includes a second preset time and / or the fluctuation of the historical accuracy of determining personnel location exceeding a preset fluctuation range. Calculate the time decay factor of the sensed data, and calculate the address confidence level corresponding to the address pointed to by the sensed data; All addresses pointed to by the sensing data in the multi-source sensing data are sequentially taken as target addresses. Based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, the comprehensive score of the target address is determined. Based on the overall score of all the target addresses, the target address with the highest overall score is selected and identified as the location of key personnel.

[0005] In some embodiments, before the step of setting the weight value of the sensing data type based on the accuracy of determining personnel location, historical average time decay factor, and average spatial correlation degree based on the history of the sensing data type, the method further includes: The accuracy rate of determining personnel location based on the number of times the historical data of the perception data type correctly determined the personnel location and the number of times the historical data of the perception data type incorrectly determined the personnel location is calculated. Based on the effective sample size and the historical time decay factor corresponding to all historical sensing data of the sensing data type, calculate the historical average time decay factor. The average spatial correlation degree is calculated based on the straight-line distance between the address and the center of the historical activity area of ​​the person in the historical sensing data of the aforementioned sensing data type.

[0006] In some embodiments, before the step of sequentially taking all addresses pointed to by the sensing data in the multi-source sensing data as target addresses, determining the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, the method further includes: Sensing data with a time decay factor less than a preset time decay factor and sensing data with an address confidence level less than a preset address confidence level are removed to update the multi-source sensing data.

[0007] In some embodiments, the method further includes: Determine whether the distance between any two addresses pointed to by the multi-source sensing data exceeds a preset distance; A data conflict group is identified, which includes any two addresses to which the sensing data points are more than a preset range apart; the distance between sensing data in different data conflict groups does not exceed the preset range. Determine the overall score of the addresses in each of the data conflict groups; The addresses in each data conflict group are ranked from highest to lowest based on their overall scores, and the address with the highest overall score in each data conflict group is determined. Sort the addresses in all data conflict groups by their highest comprehensive scores from highest to lowest, filter out the top N highest comprehensive scores, and calculate the probability of the addresses corresponding to the top N highest comprehensive scores.

[0008] In some embodiments, the calculation of the time decay factor of the sensed data corresponding to the sensed data type is performed according to the following formula: = ; in, The time decay factor is the kth sensing data corresponding to the sensing data type. The time difference is the difference between the time when the current location of the person is determined and the time when the kth sensing data is collected. This is the attenuation coefficient.

[0009] In some embodiments, the calculation of the address confidence level corresponding to the address pointed to in the sensing data is performed according to the following formula: ; in, For address; Address confidence; Address matching rate; Address standardization levels; To improve the accuracy of equipment data acquisition; These are the fusion weights for address matching rate, address standardization level, and device acquisition accuracy, respectively. .

[0010] In some embodiments, the determination of the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address is accomplished according to the following formula: ; in, For the target address; This is the overall score for the target address; The set of all perceptual data types that point to the target address; For the first Under a given perceptual data type, it is the collection of all perceptual data pointing to the target address; This is the weight value of the i-th type of sensing data corresponding to the k-th sensing data pointing to the target address; For the first Under the first perceptual data type, the first one pointing to the target address Time decay factor of sensor data; Address confidence level for the target address.

[0011] In some embodiments, the method further includes: Determine whether the location of the key personnel exceeds a preset range; When the preset range is exceeded, the current location and basic information of the key personnel are pushed to the monitoring personnel; it is determined whether the location of the key personnel is outside the district / county boundary and is in an unreported state; if the location of the key personnel is outside the district / county boundary and is in an unreported state, the current location and basic information of the key personnel are pushed to the superior unit of the monitoring personnel. The locations of key personnel determined at different times are used to form spatiotemporal trajectories, which are then visualized. Based on the spatiotemporal trajectories of the key personnel, identify their companions.

[0012] Secondly, a device for determining the location of key personnel is also provided, comprising: The first acquisition unit is used to acquire the dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types. The second acquisition unit is used to acquire the weight values ​​of the perception data types; wherein the step of determining the weight values ​​of the perception data types includes: when a preset condition is met, determining the effective sample size of historical perception data of different perception data types; if the effective sample size of historical perception data of the perception data type is lower than the corresponding preset number, then setting the weight value of the perception data type to a preset static weight value; if the effective sample size of historical perception data of the perception data type is not lower than the corresponding preset number, then setting the weight value of the perception data type based on the historical accuracy of determining personnel location, historical average time decay factor, and average spatial correlation of the perception data type; normalizing the weight values ​​of all perception data types so that the sum of all normalized weight values ​​is equal to the total number of types of all perception data types; smoothly transitioning the normalized weight values ​​with the previously obtained normalized weight values ​​to update the normalized weight values; updating the weight values ​​of the perception data types using the updated normalized weight values; the preset condition includes a second preset time and / or the fluctuation of the historical accuracy of determining personnel location exceeding a preset fluctuation range. The calculation unit is used to calculate the time decay factor of the sensed data and the address confidence level corresponding to the address pointed to by the sensed data. The first determining unit is used to sequentially take all addresses pointed to by the sensing data in the multi-source sensing data as target addresses, and determine the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address. The second determining unit is used to filter out the target address with the highest comprehensive score based on the comprehensive score of all the target addresses, and determine it as the location of key personnel.

[0013] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for determining the location of key personnel.

[0014] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method for determining the location of key personnel.

[0015] The above embodiments provide a method, apparatus, and device for determining the location of key personnel. In this method, the weights corresponding to the perceived data types are dynamically and adaptively determined based on the historical accuracy of personnel location determination, the historical average time decay factor, and the average spatial correlation. Based on the determined weights corresponding to the perceived data types, the time decay factor is fused to improve the time sensitivity of personnel location determination, and the address confidence is fused to accurately determine the comprehensive address score. The location of key personnel is then determined based on the comprehensive address score. This not only improves work efficiency and reduces personnel costs but also increases response speed. Attached Figure Description

[0016] Figure 1 An exemplary flowchart is shown for a method for determining the location of key personnel according to some embodiments; Figure 2 This illustration shows a schematic diagram of a key personnel unauthorized departure monitoring method provided in an embodiment of this application; Figure 3 This illustration shows a spatiotemporal trajectory visualization diagram provided by an embodiment of this application; Figure 4 This illustration shows a schematic diagram of spatial overlap analysis provided in an embodiment of this application; Figure 5 This illustration shows a schematic diagram of a companion identification method provided in an embodiment of this application; Figure 6 This illustration shows a schematic diagram of DBSCAN trajectory clustering provided in an embodiment of this application; Figure 7An exemplary schematic diagram of a key personnel location determination device is shown according to some embodiments. Detailed Implementation

[0017] To make the objectives and implementation methods of this application clearer, the exemplary implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings of the exemplary embodiments of this application. Obviously, the exemplary embodiments described are only some embodiments of this application, and not all embodiments.

[0018] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0019] The terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar or related objects or entities, and do not necessarily imply a specific order or sequence, unless otherwise specified. It should be understood that such terms are interchangeable where appropriate.

[0020] The terms “comprising” and “having”, and any variations thereof, are intended to cover but not exclude inclusion, for example, a product or device that includes a range of components is not necessarily limited to all of the components that are clearly listed, but may include other components that are not clearly listed or that are inherent to such product or device.

[0021] In the management of key personnel in the field of public security, the current determination of the location of key personnel still mainly relies on passive management methods such as manual visits and offline registration. This is not only inefficient and costly, but also has obvious response delays.

[0022] To address the aforementioned technical problems, embodiments of this application provide a method, apparatus, and device for determining the location of key personnel. In this method, the weights corresponding to different types of perceived data are dynamically and adaptively determined based on the historical accuracy of personnel location determination, the historical average time decay factor, and the average spatial correlation. Based on the determined weights corresponding to the perceived data types, the time decay factor is fused to improve the time sensitivity of personnel location determination, and the address confidence is also fused to accurately determine the comprehensive address score. The location of key personnel is then determined based on the comprehensive address score. This not only improves work efficiency and reduces personnel costs but also increases response speed.

[0023] Figure 1 An exemplary flowchart is shown, illustrating a method for determining the location of key personnel according to some embodiments. The method includes steps S100-S500.

[0024] S100. Obtain the dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types.

[0025] In this embodiment, the acquired personnel dynamic information includes multi-source sensing data. This avoids the problem of a single data source making it difficult to ensure data authenticity and accuracy. Because multi-source sensing data includes sensing data from different sources, the authenticity and accuracy of the data can be guaranteed, thus enabling accurate determination of the location of key personnel. Key personnel can be understood as those who require attention.

[0026] In some embodiments, the first preset time can be 5 minutes. This involves acquiring the dynamic information of key personnel over a 30-minute period.

[0027] In some embodiments, the movement information of key personnel within a first preset time period can be acquired periodically. For example, the movement information of key personnel within the first preset time period can be acquired at preset intervals. The preset interval can be 5 minutes. This allows for timely understanding of changes in the location of key personnel.

[0028] In some embodiments, the sensing data type can be face capture, vehicle capture, electronic fence, mail delivery, flight booking, flight departure, railway booking, railway security check, hotel accommodation, or internet cafe access. Correspondingly, the sensing data of the sensing data type includes face capture data, vehicle capture data, electronic fence data, mail delivery data, flight booking data, flight departure data, railway booking data, railway security check data, hotel accommodation data, or internet cafe access data. Among these, flight booking data, flight departure data, railway booking data, railway security check data, hotel accommodation data, and internet cafe access data belong to real-name tracking data. Electronic fence data includes electronic fence device data and electronic fence sensing data. It is understood that the sensing data of different sensing data types in multi-source sensing data are obtained from different data sources; for example, the sensing data of sensing data type A and the sensing data of sensing data type B have different data sources.

[0029] In some embodiments, steps S600-S1000 are included before the step of obtaining the dynamic information of key personnel within a first preset time period.

[0030] S600. Establish a multi-source sensing data extraction service and extract raw multi-source sensing data.

[0031] In some embodiments, in addition to establishing a multi-source sensing data extraction service and extracting raw multi-source sensing data, a basic data extraction service is also established and basic data is extracted. Specifically, the basic data includes key personnel data and activity location data. The steps for establishing the basic data extraction service and extracting basic data are similar to those for establishing a multi-source sensing data extraction service and extracting raw multi-source sensing data, and will not be repeated here. The process of establishing a multi-source sensing data extraction service and extracting raw multi-source sensing data is described in detail below.

[0032] In some embodiments, step S600, establishing a multi-source sensing data extraction service and extracting raw multi-source sensing data, includes steps S601-S603.

[0033] S601. Create a perception data extraction service through the timed service of the XXL-JOB (distributed task scheduling platform); XXL-JOB is a distributed task scheduling platform that connects to third-party platform APIs (Application Programming Interfaces) to independently create and configure timed extraction tasks for third-party data sources within the XXL-JOB scheduling center's management interface.

[0034] In this embodiment, a unique JobHandler (the core component in the XXL-JOB framework used to encapsulate scheduled tasks) identifier is defined for both basic data and perceived data, and a corresponding Cron (a scheduled task management tool in Linux systems) expression is set according to the data update frequency: High-frequency update: Applied to the extraction of electric field sensing data, face capture data, and vehicle capture data, it is executed every five minutes, and the expression is " The fields represent, in order: seconds (…). ),point( ),hour( ),day( ),moon( ),Week( That is, “starting from the 0th second of every minute, a task is triggered once every 5 minutes, regardless of the day, month, or weekday.”

[0035] Low-frequency updates: Applied to extract data from electronic fence devices, activity venues, key personnel, real-name tracking data, and delivery data, executed once daily, with the expression " The fields represent, in order: seconds (…). ),point( ),hour( ),day( ),moon( ),Week( This means "to trigger the task at midnight 00:00:00 every day (12:00 AM)".

[0036] Each task supports dynamic parameter passing, configured via JSON format.

[0037] S602. Control the XXL-JOB scheduling platform to trigger a timed extraction task.

[0038] In this embodiment, the XXL-JOB scheduling platform monitors the task trigger time. After obtaining the predetermined time, the scheduling platform triggers the corresponding JobHandler according to the preset Cron expression.

[0039] The scheduling platform uses the JobHandler to locate the registered executor responsible for the task and sends an HTTP request to that executor. The scheduling platform then records the results returned by the executor (success / failure, logs, etc.) in the log.

[0040] S603. Execute the extraction task to extract the raw multi-source sensing data.

[0041] When the JobHandler in the executor is scheduled and triggered, it calls the interface of the data extraction service, and then calls the third-party data source API to obtain perception data. Obtaining perception data from different third-party data source APIs achieves the extraction of raw multi-source perception data.

[0042] In this embodiment, the task scheduling adopts the XXL-JOB distributed scheduling platform to ensure the reliable execution of data extraction tasks. Using Java in conjunction with the Spring Boot framework achieves optimal implementation results. Java has a mature ecosystem in data processing, concurrent computing, and system integration, and combined with Spring Boot, it can quickly build a highly available microservice architecture. At the database level, PostgreSQL is selected in conjunction with PostGIS (Geographic Information System Extensions) spatial extensions to effectively support the storage and computation of geographic location data. Furthermore, Redis is used as a caching middleware to improve real-time query performance, and Elasticsearch is used to support fast retrieval of trajectory data, together forming a complete technology stack.

[0043] S700, Clean and standardize the sensing data in the original multi-source sensing data.

[0044] In some embodiments, the step of cleaning and standardizing the sensing data in the original multi-source sensing data includes S701-S702.

[0045] S701. Perform data parsing and format standardization on the sensing data in the original multi-source sensing data.

[0046] In this embodiment, the perception data in different formats obtained through the API is parsed to extract key field information.

[0047] For JSON data, the Jackson library is used to parse it and convert the JSON string into a Java object; for XML data, JAXB (Java Architecture for XML Binding, a technology for generating Java classes from XML Schema) is used to map XML elements to properties of Java objects, thus achieving the conversion from XML to Java objects; for CSV data, the OpenCSV library is used to parse it, read the CSV file, and map it to a Java object.

[0048] Time information from different data sources is processed using regular expressions. "The time string is matched and validated, and then uniformly converted to the ISO8601 standard format."

[0049] The Proj4J coordinate transformation library will convert geographic coordinates obtained from different data sources into the WGS-84 coordinate system used by this service, ensuring that all spatial data are calculated and displayed under the same reference.

[0050] S702. Perform quality verification and cleaning on the perceived data after data parsing and format standardization.

[0051] After data parsing and format standardization, the perceived data is checked for non-emptiness. Data with missing key business fields (such as personnel identification and time information) is directly removed. When non-key fields (such as sender's phone number) are missing, they are filled with "NULL" in a standardized manner, while other valid information is retained.

[0052] Specifically, the system uses a combination of "data source type + unique identifier + timestamp" to identify duplicate data, retaining the record with the most complete information and removing duplicate data to avoid data redundancy.

[0053] S800 stores the cleaned and standardized sensing data.

[0054] In this embodiment of the application, the cleaned and standardized sensing data will be stored in the corresponding record table in the database. Each sensing data type corresponds to one record table. The table structure consists of basic fields and business fields. Among them, face capture data, vehicle capture data and electronic fence data also contain location field information.

[0055] Basic fields are typically fields related to the fundamental information of the data record itself, such as unique identifier, creation time, update time, and data source; business fields are fields related to the specific business logic perceived by personnel, such as personnel's name, ID number, and various business statuses and descriptions; location fields are fields directly related to geographical location, such as longitude and latitude. Table 1 lists the key business fields and location fields in various types of perceived data record tables.

[0056] Table 1

[0057] Among them: mobile phone MAC address (Media Access Control Address); mobile phone IMSI address (International Mobile Subscriber Identity).

[0058] Each of the above record tables includes perception data for all individuals. For example, the internet cafe internet access record table includes perception data of all individuals' internet cafe usage, which is a type of perception data.

[0059] S900 performs maintenance processing on the cleaned and standardized sensor data stored.

[0060] Specifically, maintaining and processing the cleaned and standardized sensing data includes periodically backing up the cleaned and standardized sensing data, periodically deleting the cleaned and standardized sensing data, and establishing a database index for the database storing the cleaned and standardized sensing data.

[0061] The cleaned and standardized sensor data is backed up regularly using a dual-mode backup strategy combining full and incremental backups. For example, a full backup is performed weekly, while multiple incremental backups are performed daily during off-peak hours (such as early morning). Backup data is stored on different media tiers based on its importance and access frequency. Recent critical backups are stored on high-speed disk arrays for rapid recovery, while historical archive backups are automatically transferred to lower-cost object storage or tape libraries. After each backup task, the service automatically verifies, calculates, and performs recovery tests on the backup files to ensure the validity and recoverability of the backup data.

[0062] The scheduled deletion of the cleaned and standardized sensor data is specifically achieved by utilizing the distributed task scheduling framework XXL-JOB to manage and execute periodic data cleanup tasks. Within the XXL-JOB framework, the system runs pre-written data cleanup scripts. These scripts automatically perform deletion operations based on the administrator's preset data retention policy (e.g., retaining raw data for 6 months and intermediate results for 1 year). XXL-JOB supports task fragmentation and broadcasting, distributing massive data cleanup tasks to multiple executor nodes for parallel processing, significantly improving cleanup efficiency.

[0063] The cleanup script supports precise filtering by multiple dimensions such as data type, data time, and associated personnel ID. Before performing a deletion operation, the system first performs logical deletion (soft deletion), marking the data as "pending deletion" and transferring it to a temporary area. After confirming that there is no business impact, the script then performs physical deletion.

[0064] The database indexing of the cleaned and standardized sensor data was specifically optimized to address the numerous real-time queries and historical trajectory analyses involved in location determination, significantly improving data retrieval speed and system responsiveness. Highly efficient composite indexes were created for core query conditions. For example, a composite index with (personnel ID, data time) as the key was created on the sensor data record table, optimizing the efficiency of trajectory queries by personnel and time range. For data containing geographic coordinates, spatial database indexes (such as R-Tree indexes, another form of B-tree development towards multi-dimensional space) were used to support rapid response for spatial range queries.

[0065] The built-in index usage monitoring module regularly analyzes the effectiveness of indexes. It cleans up redundant indexes with low usage or those affecting write performance, and dynamically suggests or creates new indexes for new high-frequency query conditions, achieving adaptive optimization of the indexing strategy.

[0066] After establishing a database index for the database storing cleaned and standardized sensing data, the sensing data can be searched by personnel and time range.

[0067] S1000 performs location data fusion on the cleaned and standardized sensing data stored in the system.

[0068] The steps for fusion of location data into the cleaned and standardized perception data include: unifying identity identification association mapping and generating a standardized dataset, and generating a dynamic queryable data table of personnel by combining a data indexing mechanism.

[0069] The unified identity association mapping is used to construct a personnel identity alignment model across data elements. By using association rule mining and information complementarity mechanisms, it deeply integrates identity association clues in multi-source datasets to generate globally unique and stable personnel identity identifiers, providing a core index for connecting personnel trajectories across data.

[0070] The standardized dataset was generated by comprehensively integrating personnel trajectory information from different sensing data systems and business systems, based on the alignment of spatiotemporal benchmarks with personnel identification. Through the fusion of multi-source heterogeneous data, a unified data view centered on personnel identification was formed, thus constructing a standardized dynamic personnel dataset. Specifically, it includes fields such as personnel name, personnel identity, mobile phone number, ID card number, sensing device name / sensing method, sensing address, sensing device type, location number, IMSI, associated device number, associated record table ID, dynamic information description, and record occurrence time.

[0071] Step S1000 involves splitting the perception data record table in the database according to personnel identification, resulting in perception data record tables corresponding to different personnel identifications. Each perception data record table corresponding to a personnel identification record contains all perception data of all perception data types for the personnel corresponding to that identification.

[0072] By combining a data indexing mechanism to generate a dynamically queryable personnel data table, this feature provides a combined query method for multi-dimensional attributes such as personnel identity, time range, and spatial location. It can quickly retrieve and dynamically generate dynamic data tables for key personnel within any time period. The final generated dynamically queryable personnel data table allows users to search for the dynamic information of a specific person within a preset time period.

[0073] S200, Obtain the weight value of the perceived data type.

[0074] In this embodiment of the application, the weight value of each type of sensing data may be different. In this embodiment of the application, the weight value of the sensing data type is determined based on the historical sensing data. In this embodiment of the application, the step of determining the weight value of the sensing data type includes S201-S205.

[0075] S201. When the preset conditions are met, determine the effective sample size of historical sensing data of different sensing data types.

[0076] In this embodiment of the application, the effective sample size of historical sensing data of different sensing data types refers to the number of effective samples in historical sensing data of different sensing data types. Effective samples refer to historical sensing data that can be used to determine the location of key personnel.

[0077] For example, the effective sample size of historical perceptual data of perceptual data type A is a', and the effective sample size of historical perceptual data of perceptual data type B is b'.

[0078] In this embodiment, the weight value of the perceived data type is continuously updated based on historically obtained perceived data, and an update is required when a preset condition is met. The preset condition includes a second preset time and / or the fluctuation of the accuracy of historically determined personnel location exceeding a preset fluctuation range.

[0079] In one example, the second preset time can be daily or weekly. Understandably, daily updates are preferable in scenarios with high staff turnover and frequent data updates, while weekly updates are preferable in scenarios with stable staff activity patterns.

[0080] The weight value of a certain perception data type can be updated when the accuracy of historical personnel location determination fluctuates beyond a preset fluctuation range. In one example, the preset fluctuation range is ±10%.

[0081] In this embodiment, the weight iteration cycle can be flexibly configured according to the characteristics of the business scenario. It can be updated daily (suitable for scenarios with high personnel mobility and frequent data updates), updated weekly (suitable for scenarios with stable personnel activity patterns), or triggered iteration (updated immediately when the accuracy of a certain type of perception data fluctuates by more than ±10%).

[0082] S202. If the effective sample size of the historical sensing data of the sensing data type is lower than the corresponding preset number, then the weight value of the sensing data type is set to a preset static weight value.

[0083] In this embodiment, different sensing data types are all within corresponding preset quantities, which can be specifically set according to the sensing data type. For example, when the effective sample size of historical sensing data of sensing data type A is lower than the corresponding preset quantity (e.g., 30), the weight value of the sensing data type is set to a preset static weight value. Similarly, when the effective sample size of historical sensing data of sensing data type B is lower than the corresponding preset quantity (e.g., 50), the weight value of the sensing data type is set to a preset static weight value.

[0084] For example, the credibility of ten types of sensing data, including face capture data, vehicle capture data, electronic fence sensing data, postal information data, flight ticketing data, flight departure data, railway ticketing data, railway inspection data, hotel accommodation data, and internet cafe browsing records, is established using a static weight system with specific location determination, as shown in Table 2.

[0085] Table 2

[0086] S203. If the effective sample size of the historical sensing data of the sensing data type is not less than the corresponding preset number, then the weight value of the sensing data type is set according to the accuracy of determining the location of personnel, the historical average time decay factor and the average spatial correlation of the historical sensing data type.

[0087] In this embodiment of the application, since there are a large number of effective samples, the weight values ​​of the sensing data types can be dynamically set according to the historical sensing data, so that the weight values ​​set in this way are more accurate.

[0088] In some embodiments, the weight value of the perceived data type, based on the accuracy of determining personnel location based on the history, the historical average time decay factor, and the average spatial correlation, can be calculated using the following formula: ; in, Let be the weight value of the sensing data corresponding to the i-th sensing data type; The accuracy of determining personnel location for the history of the i-th type of sensory data; is the average time decay factor corresponding to the i-th sensing data type; Let be the average spatial correlation of the i-th perceptual data type; The weights for the accuracy of historical personnel location determination, the average time decay factor, and the average spatial correlation degree are respectively required to meet the following conditions. Normalization constraints. Generally, empirical configuration is used. It is 0.7. It is 0.2. It is 0.1, but it can be adjusted according to specific business scenarios.

[0089] The accuracy of historical personnel location determination reflects the accuracy of the perceived data in determining personnel location. The average time decay factor reflects the average overall timeliness of the perceived data, with a value range of [0, 1], where the closer to 1, the stronger the timeliness. The average spatial correlation reflects the average spatial correlation between the address pointed to by the perceived data and the historical frequently active area of ​​the personnel. The address within the historical frequently active area of ​​the personnel is defined as 1, and decays with increasing distance, with a value range of [0, 1].

[0090] In this embodiment, a dynamic adaptive data weight adjustment mechanism is constructed, with the core objective of overcoming the limitations of traditional fixed and static weight configurations. This mechanism uses the accuracy of historical data types as a benchmark, integrates data timeliness and spatial correlation characteristics, and dynamically optimizes the weight contribution of various types of perceived data, ultimately achieving adaptive control of the weights corresponding to different types of perceived data. The overall design adopts a three-layer architecture to achieve intelligent full-cycle weight management: Basic layer: Accuracy determined historically. As the core anchor point, an initial weight benchmark is established for various types of perceived data to ensure that the weights are bound to the reliability of the data itself; Correction layer: business characteristics such as data timeliness and spatial correlation are introduced to finely calibrate the benchmark weights; Iteration layer: a time-driven closed-loop feedback system is built to continuously learn and iterate to optimize the weight parameters by periodically (e.g., daily / weekly) and by analyzing the fluctuations in the accuracy of determining personnel locations in history, so as to achieve adaptive adjustment of weights to data quality fluctuations.

[0091] In some embodiments, before the step of setting the weight value of the sensing data type based on the accuracy of determining personnel location, historical average time decay factor, and average spatial correlation degree based on the history of the sensing data type, the method further includes: The accuracy rate of determining personnel location based on the number of times the historical data of the perception data type correctly determined the personnel location and the number of times the historical data of the perception data type incorrectly determined the personnel location is calculated. Specifically, the following formula can be used for calculation: ; in, The accuracy of determining personnel location for the history of the i-th type of sensory data; The number of times the location of a person was correctly determined for the history of the i-th type of sensory data; The number of times the personnel location is determined for the historical error of the i-th perception data type.

[0092] The accuracy rate of determining the location of personnel based on the historical data of the i-th type of sensing data can be understood as the accuracy rate of determining the location of personnel based on the historical sensing data of the i-th type of sensing data. Other parameters in the above formula can be deduced similarly.

[0093] Based on the effective sample size and the historical time decay factor corresponding to all historical sensing data of the sensing data type, calculate the historical average time decay factor. Specifically, the following formula can be used for calculation: ; in, is the average time decay factor corresponding to the i-th sensing data type; The effective sample size of historical sensing data for the i-th sensing data type; The attenuation coefficient; This is the difference in acquisition time (in minutes) between the kth historical sensing data and the kth historical sensing data of the i-th sensing data type.

[0094] The average spatial correlation degree is calculated based on the straight-line distance between the address and the center of the historical activity area of ​​the person in the historical sensing data of the aforementioned sensing data type.

[0095] Specifically, the following formula can be used for calculation: ; in, Let be the average spatial correlation of the i-th perceptual data type; The effective sample size of historical sensing data for the i-th sensing data type; For the i-th perceptual data type, the th The address pointed to by the historical sensing data Center of personnel history and frequent activity areas The straight-line distance between latitude and longitude; ; In the above formula, 111 represents the approximate surface distance (unit: km) corresponding to each degree of latitude and longitude. Latitude; Longitude. This represents the difference in latitude between the address and the center of activity. This indicates the difference in longitude between the address and the center of activity. It is the average of the latitude of the two points (address and center of activity). As a correction factor, it is used to correct the phenomenon that the distance in the longitude direction decreases as latitude increases.

[0096] S204. Normalize the weight values ​​of all the sensory data types so that the sum of all normalized weight values ​​equals the total number of types of all the sensory data types.

[0097] In this embodiment, the weight values ​​of different sensing data types may vary significantly, and direct use would lead to an imbalance in weight allocation. Therefore, normalization is required. Specifically, the weight values ​​are scaled proportionally so that the sum of the normalized weight values ​​of all sensing data types equals the total number of types of the sensing data types. This process maintains the relative importance ratio between weights while calibrating their sum to a fixed total number of types, thus achieving weight standardization and balance.

[0098] Specifically, normalization can be performed using the following formula: ; in, The normalized weight value for the i-th perceptual data type; Let be the weight value of the sensing data corresponding to the i-th sensing data type; Let be the perceptual data weight for the j-th perceptual data type, 1≤j≤ ; This represents the total number of types across all perceptual data types.

[0099] S205. The normalized weight value and the previously obtained normalized weight value are smoothly transitioned to update the normalized weight value; the updated normalized weight value is used to update the weight value of the perceptual data type.

[0100] In this embodiment of the application, after executing S205, the weight value of the current perceptual data type is the weight value after the normalization process.

[0101] In this embodiment, to avoid large fluctuations in the location determination results caused by sudden changes in weight values ​​and to ensure the stability and continuity of the determination results, a moving average method is used to achieve a smooth transition between the old and new weights. The smooth transition can be accomplished according to the following formula: ; in, It is the updated normalized weight value of the i-th perceptual data type; It is the smoothing coefficient. The smaller the value, the more smoothly the weight parameter changes; It is the weight value of the normalized processing of the i-th perceptual data type; It is the weight value that took effect in the previous iteration of the i-th perceptual data type, that is, the weight value of the normalization process obtained in the previous iteration of the i-th perceptual data type.

[0102] S300, calculate the time decay factor of the sensed data, and calculate the address confidence level corresponding to the address pointed to by the sensed data.

[0103] In this embodiment, a time decay function is introduced to evaluate the reference value of the sensing data based on the acquisition time, ensuring that the most up-to-date data receives higher weight in the analysis.

[0104] Time decay factor ( This is used to quantify the degree of decay in the reference value of a single piece of sensing data over time. Its value ranges from (0,1), and the closer the value is to 1, the stronger the timeliness of the data.

[0105] In some embodiments, the calculation of the time decay factor of the sensed data is performed according to the following formula: = ; in: The time decay factor is the kth sensing data corresponding to the sensing data type. The time difference is the difference between the time when the current location of the person is determined and the time when the kth sensing data is collected (in minutes). This is the attenuation coefficient, which can be dynamically adjusted according to specific business scenarios.

[0106] In this embodiment, each type of sensing data may correspond to multiple sensing data points, and each sensing data point has a corresponding time decay factor. In some implementations, a lower limit constraint on the time decay factor can also be set: ≥0.01. When the calculated time decay factor is lower than 0.01, it is uniformly set to 0.01 to avoid the weight returning to zero.

[0107] Taking a typical control scenario (λ=0.02) as an example, the values ​​corresponding to different time differences are shown in Table 3 below.

[0108] Table 3

[0109] In this embodiment of the application, in order to overcome the limitations of the traditional fixed coefficient configuration of address confidence, the application relies on the real-time changes of address data quality, collection context features, and historical determination results, that is, dynamically optimizes the fusion weight ratio of address matching rate, address standardization level, and device collection accuracy, and finally realizes intelligent adaptive adjustment of address confidence, supporting the improvement of the accuracy of location determination algorithm.

[0110] In some embodiments, the calculation of the address confidence level corresponding to the address pointed to in the sensing data is performed according to the following formula: ; in, For address; Address confidence; Address matching rate; Address standardization levels; To improve the accuracy of equipment data acquisition; These are the fusion weights for address matching rate, address standardization level, and device acquisition accuracy, respectively. .

[0111] in addition, ; The total number of times address h has appeared in history. This represents the number of times the address has been determined to be a correct match. A correct match can be understood as the person's actual location being at that address. This is also the number of valid historical records for that address. At that time, the Laplace smoothing algorithm was used to correct the matching rate. If there are no historical records, industry prior values ​​are used to avoid interference from small sample statistical bias in the matching rate calculation.

[0112] In this embodiment of the application, a fusion coefficient is set. These are the fusion weights for address matching rate, address standardization level, and device acquisition accuracy, respectively, to meet the following requirements. Normalization constraints. The default configuration is generally (0.6, 0.2, 0.2), but it can be adjusted according to the priority of specific business scenarios.

[0113] In the embodiments of this application, Address matching rate reflects the reliability of matching historical records of an address with the actual location of a person. Address standardization level quantifies the precision of the address description, with a value range of [0, 1]. For example, when the precision is down to a specific longitude and latitude, rather than a range, the address standardization level is higher, and the corresponding value is higher. The acquisition accuracy of the device reflects the inherent accuracy level of the device or system that acquires the data at this address, and its value range is [0, 1].

[0114] In this embodiment, address matching rate is used as the basis for measurement. An initial benchmark for coefficients is established using anchor points. Fine-grained feature benchmark libraries are constructed based on the degree of address standardization and the device accuracy of the data source, respectively, to dynamically fine-tune the weights of the base layer. In some embodiments, confidence assessment results can be periodically compared with actual business verification results, using the difference signals to drive automated iterative learning of the weight parameters.

[0115] S400. Take all addresses pointed to by the sensing data in the multi-source sensing data as target addresses in sequence. Based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, determine the comprehensive score of the target address.

[0116] For example, in the multi-source sensing data, all addresses pointed to by the sensing data include Y and U. Addresses Y and U can be used as target addresses, and a comprehensive score for each target address is determined by applying a weight value based on the sensing data type corresponding to the sensing data pointing to the target address, a time decay factor for the sensing data pointing to the target address, and an address confidence level for the target address. In some embodiments, the determination of the comprehensive score for the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor for the sensing data pointing to the target address, and the address confidence level for the target address is performed according to the following formula: ; in, For the target address; This is the overall score for the target address; The set of all perceptual data types that point to the target address; For the first Under a given perceptual data type, it is the collection of all perceptual data pointing to the target address; This is the weight value of the i-th type of sensing data corresponding to the k-th sensing data pointing to the target address; For the first Under the first perceptual data type, the first one pointing to the target address Time decay factor of sensor data; Address confidence level for the target address.

[0117] For example, when using address Y as the target address, the first step is to determine the set of all perceptual data types that point to address Y. This set includes two sensor data types: the first is face capture, and the second is vehicle capture. Determine the set of all sensor data pointing to address Y under the first sensor data type. And, under the second type of perceptual data, the set of all perceptual data pointing to address Y. .

[0118] Under the first type of perceptual data, it is the set of all perceptual data pointing to address Y. This includes two sets of sensing data. Specifically, under the first sensing data type, the weight value of the sensing data type corresponding to the first sensing data pointing to address Y is determined, and under the first sensing data type, the weight value of the sensing data type corresponding to the second sensing data pointing to address Y is determined. The weight value of the same sensing data type is always the same, which is the weight value corresponding to the first sensing data type; for example, both are 0.9. Additionally, the time decay factor of the first sensing data pointing to address Y under the first sensing data type is determined; for example, it is 0.3. The time decay factor of the second sensing data pointing to address Y under the first sensing data type is determined; for example, it is 0.5.

[0119] Under the second type of perceptual data, it is the set of all perceptual data pointing to address Y. This includes two types of sensing data. Specifically, under the second sensing data type, the weight value of the sensing data type corresponding to the first sensing data pointing to address Y is determined, and under the second sensing data type, the weight value of the sensing data type corresponding to the second sensing data pointing to address Y is determined. The weight value of the same sensing data type is always the same, which is the weight value corresponding to the second sensing data type; for example, both are 0.7. Additionally, the time decay factor of the first sensing data pointing to address Y under the second sensing data type is determined; for example, it is 0.4. The time decay factor of the second sensing data pointing to address Y under the second sensing data type is determined; for example, it is 0.9.

[0120] The address confidence level for address Y is determined to be 0.8.

[0121] At this point, the total score for address Y includes 0.9×0.3×0.8+0.9×0.5×0.8+0.7×0.4×0.8+0.7×0.9×0.8=1.304.

[0122] In this embodiment, three core elements—sensing data type weight, time decay, and address confidence—are integrated to classify, weight, and fuse multi-source sensing data, ultimately achieving a refined and robust assessment of the possible locations of personnel. It can calculate the comprehensive score of the target address pointed to by sensing data within a first preset time period.

[0123] In some embodiments, before the step of sequentially taking all addresses pointed to by the sensing data in the multi-source sensing data as target addresses, determining the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, the method further includes: Sensing data with a time decay factor less than a preset time decay factor and sensing data with an address confidence level less than a preset address confidence level are removed to update the multi-source sensing data.

[0124] For example, the preset time decay factor can be 0.1. Sensing data with a time decay factor less than 0.1 is discarded. Sensing data with a time decay factor less than 0.1 means it has expired and is of little significance for determining personnel location.

[0125] The preset address confidence level can be 0.3. Sensing data with an address confidence level less than 0.3 is discarded. Sensing data with an address confidence level less than 0.3 means that the address pointed to by the sensing data is not very reliable and is not very meaningful for determining the location of people.

[0126] S500. Based on the comprehensive score of all the target addresses, select the target address with the highest comprehensive score and determine it as the location of key personnel.

[0127] In this embodiment of the application, the target address with the highest comprehensive score is determined as the location of key personnel.

[0128] In some embodiments, the method further includes: Determine whether the distance between any two addresses pointed to by the multi-source sensing data exceeds a preset distance.

[0129] In some embodiments, the step of determining whether the distance between any two addresses pointed to by the sensing data exceeds a preset distance includes: Determine the region where the address is located; If the areas of the addresses pointed to by the two sets of sensing data span cities, then it is determined whether the distance between the addresses pointed to by the two sets of sensing data exceeds a first preset distance. For example, the first preset distance is 20km.

[0130] If the areas where the addresses pointed to by the two sets of sensing data do not cross a city, and the area where either address is located is in a suburban area, then it is determined whether the distance between the addresses pointed to by the two sets of sensing data exceeds a second preset distance. For example, the second preset distance is 5km.

[0131] If the addresses pointed to by the two sets of sensing data do not cross city boundaries and are both located in the city's core area, then it is determined whether the distance between the addresses pointed to by the two sets of sensing data exceeds a third preset distance. For example, the third preset distance is 3km.

[0132] A data conflict group is determined, which includes any two addresses to which the sensing data points are more than a preset range apart; the distance between sensing data in different data conflict groups does not exceed a preset range.

[0133] In one example, data collision group A includes addresses a, b, and c. Data collision group B includes addresses d and e.

[0134] Determine the overall score of the addresses in each of the data conflict groups.

[0135] In this embodiment of the application, the process of calculating the comprehensive score of the address in the data conflict group is the same as the process of determining the comprehensive score of the target address described above, and will not be repeated here.

[0136] For example, the comprehensive scores of addresses a, b, and c in data conflict group A are determined, which can be 80, 90, and 98 respectively. The comprehensive scores of addresses d and e in data conflict group B are determined, which can be 40 and 60 respectively.

[0137] The addresses in each data conflict group are ranked from highest to lowest based on their overall scores, and the address with the highest overall score in each data conflict group is determined.

[0138] In one example, the addresses in data conflict group A are ranked by their overall scores, and the address with the highest overall score in data conflict group A is identified as address b. Similarly, the addresses in data conflict group B are ranked by their overall scores, and the address with the highest overall score in data conflict group B is identified as address e.

[0139] Sort the addresses in all data conflict groups by their highest comprehensive scores from highest to lowest, filter out the top N highest comprehensive scores, and calculate the probability of the addresses corresponding to the top N highest comprehensive scores.

[0140] For example, N can be 3. For example, the probabilities of addresses b and e need to be calculated.

[0141] In some embodiments, the probability of an event can be calculated using the following formula: probability ; in, Addresses in the set of valid candidate addresses; For the set of valid candidate addresses For any address in the list, form a set of valid candidate addresses by combining the addresses corresponding to the highest comprehensive scores of the top N positions. The probability of an address; This represents the sum of the combined scores of all addresses in the set of valid candidate addresses.

[0142] In this embodiment, the comprehensive scores of all unfiltered valid addresses are normalized to obtain the probability of each address. Calculating the probability of the address corresponding to the top N highest comprehensive scores can provide the public security system with a more intuitive probability of determining a person's location, making it easier to use in practice.

[0143] In some embodiments, the method further includes: Determine whether the location of the key personnel exceeds a preset range; When the location exceeds a preset range, the current location and basic information of the key personnel are pushed to the monitoring personnel; it is then determined whether the location of the key personnel is outside the district / county boundary and in an unreported state; if the location of the key personnel is outside the district / county boundary and in an unreported state, the current location and basic information of the key personnel are pushed to the monitoring personnel's superior unit. In one example, the preset range is an electronic fence area.

[0144] In the embodiments of this application, such as Figure 2 As shown, the system uses a high-precision map to define an electronic fence area, i.e., a preset range, to monitor and issue early warnings in real time for the activity trajectories of key personnel. If a person leaves the designated area without permission, the system immediately triggers an alert and pushes the person's basic information and current location in real time. Furthermore, if the system determines through a location determination algorithm that the person has crossed a district / county boundary and is in an unreported state, the alert level will automatically be upgraded to a severe level, and the relevant information will be simultaneously and directly pushed to the superior authority.

[0145] The locations of key personnel determined at different times are used to form spatiotemporal trajectories, which are then visualized. In this embodiment, the location information generated by the location determination algorithm is sorted and linearized according to chronological order to form and visualize a spatiotemporal trajectory. An interactive timeline is used to visually annotate key events and add detailed notes, clearly presenting the location and status of key personnel at different points in time.

[0146] Simultaneously, the trajectory information is geocoded into latitude and longitude coordinates. Based on a 3D map scene, precise mapping between dynamic personnel data and geospatial coordinates is achieved. Then, waypoints are integrated according to time sequence to generate continuous geographical paths, accurately reconstructing the complete travel trajectories of key personnel. Figure 3 As shown. Figure 3 A single data trajectory is a trajectory obtained based on one type of sensing data. A fused data trajectory is a trajectory obtained based on multi-source sensing data.

[0147] Based on the spatiotemporal trajectories of the key personnel, identify their companions.

[0148] In this embodiment, the risk of clustering caused by key personnel traveling together is dynamically determined. Through spatiotemporal fusion analysis of multi-source data, the accompanying trajectory of key personnel in the same time period and the same area is accurately captured, realizing comprehensive coverage and in-depth analysis of personnel movement trajectories across regions and scenarios, and significantly improving the precision level and forward-looking early warning capability of public safety prevention and control.

[0149] In some embodiments, the step of identifying fellow travelers based on the spatiotemporal trajectories of the key personnel includes: setting query conditions; performing spatiotemporal overlap analysis based on the query conditions to determine candidate fellow travelers; and using LCSS calculation to determine the fellow travelers by combining the spatiotemporal trajectories of the candidate fellow travelers and the spatiotemporal trajectories of the key personnel.

[0150] Set query criteria: Select a key person as the analysis target, determine the time range (e.g., within 24 hours) and spatial range (e.g., within a district or county), and ensure that the activity cycle of the person is covered, and complete the preliminary screening of all relevant trajectory data within the time and spatial ranges.

[0151] Spatiotemporal overlap analysis is performed based on the query conditions: within the time and space range, other personnel who highly overlap with the key personnel in time and space are identified based on a preset spatial threshold (e.g., 500 meters) and time window (e.g., 10 minutes) and determined as candidate companions.

[0152] Specifically, if the location is within the spatial threshold, it is considered "spatial overlap"; if it occurs within the same time window, it is considered "temporal overlap"; if spatiotemporal overlap is detected at multiple consecutive time points, i.e., both spatial and temporal overlap, then multiple individuals who have spatiotemporal overlap with the key personnel will be listed as candidate travel companions. Figure 4 As shown, trajectory 1 is the spatiotemporal trajectory of key personnel, and trajectory 2 is the spatiotemporal trajectory of other personnel.

[0153] ; in, and These represent arbitrary trajectory points of the two individuals. For the preset spatial threshold, This is a preset time window. s|represents the trajectory point and The planar Euclidean distance between them, | t represents the time difference between two points. p represents key personnel, and q represents other personnel; The latitude of key personnel locations. The latitude of the other personnel's locations; The longitude of the location of key personnel. The longitude of the location of other personnel. This refers to the time when key personnel appeared at the location of key personnel; This refers to the time when other personnel appear in the locations of other personnel.

[0154] The Longest Common Subsequence (LCSS) algorithm is used to find common subsequences with consistent spatiotemporal order between the spatiotemporal trajectories of candidate peers and key personnel. Based on dynamic programming principles, the trajectory similarity score between the key personnel and other candidate peers is calculated. When the similarity between the two trajectories exceeds 70%, they are identified as peers; when the similarity is between 50% and 70%, they are identified as suspected peers. Figure 5 As shown.

[0155] ; ; in, The longest common subsequence that satisfies spatiotemporal similarity between the trajectory sequences of key individuals and the trajectory sequences of candidates; the trajectory sequence of key individuals and candidate trajectory sequence The lengths are respectively and ; Trajectory sequence The There are several trajectory points, whose spatiotemporal coordinates are: , Represents the latitude of the c-th trajectory point. This represents the longitude of the c-th trajectory point. This represents the occurrence time corresponding to the c-th trajectory point; Trajectory sequence The There are several trajectory points, whose spatiotemporal coordinates are: , Represents latitude, Represents longitude. For the time of occurrence, Represents the latitude of the d-th trajectory point. This represents the longitude of the d-th trajectory point. This represents the occurrence time corresponding to the d-th trajectory point; the parameter value range is... , ; The method is used to calculate the spatial distance between two trajectory points. The preset similarity threshold for trajectory points; The method returns the original length of the sequence. This represents the original length of trajectory sequence A.

[0156] Output results: After identifying fellow travelers, a sequence of key locations ordered by time is automatically generated, accurately outlining the traveler's route. Simultaneously, matching common trajectory segments are highlighted in a visual map, while unmatched segments are grayed out, intuitively reconstructing the entire journey of the traveler.

[0157] In some embodiments, the method further includes determining the spatial distribution, movement scale, and dynamic flow direction of the population.

[0158] In this embodiment of the application, in response to potential safety hazards such as excessive crowd gathering and association with high-risk areas, a large amount of personnel trajectory data is aggregated and deeply mined in real time to accurately grasp the spatial distribution, movement scale and dynamic flow of the crowd, thereby achieving real-time perception of the overall group situation and accurate assessment of potential risks.

[0159] Trajectory vectorization: For each key personnel's movement trajectory within a specific time and space range, the key coordinate points on the trajectory are accurately extracted and then transformed into a series of directed trajectory segments that are connected end to end, thereby forming structured continuous trajectory data.

[0160] Route density clustering: Using the DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise), unsupervised clustering analysis is performed on all personnel trajectory data, providing an adjustable neighborhood radius. ) and minimum number of trajectories ( The parameters are used to find high-density similar path regions in the trajectory space and accurately capture movement patterns with group characteristics.

[0161] 1) The DBSCAN algorithm defines the cluster density using two core parameters: Neighborhood radius ( ): Defines the neighborhood range of the trajectory line. If the distance between trajectory line A and trajectory line B is less than 1 / 2... If the neighborhood is defined, then trajectory line A is called a neighbor of trajectory line B.

[0162] Minimum number of samples ( A trajectory line Within the neighborhood, it must contain at least Only a single trajectory line can be considered a core trajectory line.

[0163] 2) The workflow of the DBSCAN algorithm is as follows: Calculate the core trajectory line: Randomly select an unvisited trajectory line from the trajectory set and calculate its... The number of trajectory lines within the neighborhood. If the number is greater than or equal to... If so, the trajectory line is marked as the core trajectory line.

[0164] Extended clustering: For each core trajectory line, it is... Trajectory lines within the neighborhood are added to the current cluster; if these added trajectory lines are also core trajectory lines, their neighborhoods are recursively added to the cluster. Trajectory lines within the neighborhood continue to be added to the current cluster until no further expansion is possible.

[0165] Noise trajectory lines: Trajectories that are not included in any cluster are marked as noise trajectories.

[0166] Repeat the process: Select unvisited trajectory lines from the trajectory set and repeat the above three steps until all trajectory lines have been visited, at which point the clustering process ends. Figure 6 As shown, Figure 6 It includes cluster 1, cluster 2, and noise trajectories.

[0167] Visualized situation map: An intuitive mobile situation visualization interface is built on the system map. High-frequency paths are identified and rendered in the form of dynamic flow lines. The color intensity is positively correlated with the frequency of path use and the intensity of personnel flow.

[0168] Key node annotation: Key nodes discovered during the aggregation process are overlaid on the same map view as marker points, comprehensively presenting the flow of people and hubs.

[0169] The method in this application embodiment addresses the actual needs of public security and security personnel management. Based on multi-source sensing data and centered on accurate location determination of key personnel, it is guided by practical application. Through the coordinated linkage of seven key links, namely "data extraction, data processing, data storage, data maintenance, dynamic fusion, location determination, and comprehensive application", it achieves accurate grasp of personnel dynamics, completes precise mapping of location information and efficient implementation of management decisions, and fully supports the management of key personnel under the background of smart policing.

[0170] In this embodiment, basic data such as key personnel data and activity location data are first aggregated, along with perceived data such as facial capture data, vehicle capture data, electronic fence data, postal data, and real-name trajectory data, breaking the limitations of a single data source. The data processing stage completes data parsing, format conversion, data verification, and outlier removal to ensure data quality and consistency. The data storage stage achieves secure storage and unified retrieval of massive amounts of data. The data maintenance stage utilizes regular backups, periodic deletions, and database indexing mechanisms to maintain data availability. Then, by associating personnel identification with modeling, the aggregation and fusion of multi-source data are achieved, constructing a dynamic data view centered on personnel identity. Based on this, a dynamic adaptive weighting system for perceived data is established, relying on quantitative assessment of data credibility, and combined with time decay factors and spatial correlation calculations to effectively improve the accuracy of location information acquisition. Finally, in the comprehensive application stage, based on accurate location determination results, elements such as abnormal personnel trajectories, companion trajectories, and group dynamics are deeply mined to achieve dynamic risk warning and proactive control of key personnel.

[0171] This invention constructs a complete technical solution for determining the location of key personnel. Based on multi-source sensing data resources in the social sphere, it achieves automated extraction and cleaning of multi-source heterogeneous sensing data through distributed task scheduling. It innovatively establishes an adaptive enhanced weighting system based on the credibility of sensing data, and introduces a time decay factor and address confidence level to map multi-source data to a unified specific geographical location. Finally, it generates accurate location determination results based on a weighted probability fusion algorithm. The purpose of this invention is to solve the technical problems of traditional methods, such as single data source, limited applicable groups, insufficient location determination accuracy, poor real-time performance, and limited practicality of results. By transforming multi-source data into probabilistic determination results at the specific address level, it achieves a leap in location accuracy from kilometer-level grid to specific addresses, significantly improving the determination accuracy and providing minute-level update capabilities, thus providing direct and usable technical support for precise public safety management.

[0172] The key aspect of the method in this application is the construction of a complete mapping and calculation system from multi-source sensing data to precise addresses. Specific protection points include: First, based on historical data types, determining accuracy, average timeliness characteristics, and spatial correlation characteristics of the effect association, constructing a dynamic adaptive sensing data weighting system, particularly developing differentiated weight allocation schemes for ten types of sensing data; Second, a dynamic weight adjustment mechanism incorporating time decay factors to ensure the time sensitivity of the determination results; Third, dynamically optimizing the fusion weight ratio of address matching rate, address standardization level, and device acquisition accuracy to achieve intelligent control of address confidence; Fourth, a precise address calculation model based on weighted probability fusion to achieve probabilistic determination from multi-source data to specific addresses; Fifth, an output mechanism oriented towards practical applications, capable of directly outputting a probability-sorted list of specific addresses and corresponding data source analysis.

[0173] The above embodiments provide a method, apparatus, and device for determining the location of key personnel. In this method, the weights corresponding to the perceived data types are dynamically and adaptively determined based on the historical accuracy of personnel location determination, the historical average time decay factor, and the average spatial correlation. Based on the determined weights corresponding to the perceived data types, the time decay factor is fused to improve the time sensitivity of personnel location determination, and the address confidence is fused to accurately determine the comprehensive address score. The location of key personnel is determined based on the comprehensive address score. This not only improves work efficiency and reduces personnel costs but also increases response speed. Furthermore, as... Figure 1 In terms of specific implementation of the method, this application provides a device for determining the location of key personnel, such as... Figure 7 As shown, it includes: The first acquisition unit 701 is used to acquire the dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types. The second acquisition unit 702 is used to acquire the weight value of the perception data type; wherein the step of determining the weight value of the perception data type includes: when a preset condition is met, determining the effective sample size of historical perception data of different perception data types; if the effective sample size of historical perception data of the perception data type is lower than the corresponding preset number, then setting the weight value of the perception data type to a preset static weight value; if the effective sample size of historical perception data of the perception data type is not lower than the corresponding preset number, then setting the weight value of the perception data type based on the historical accuracy of determining personnel location, historical average time decay factor, and average spatial correlation of the perception data type; normalizing the weight values ​​of all perception data types so that the sum of all normalized weight values ​​is equal to the total number of types of all perception data types; smoothly transitioning the normalized weight value to the previously obtained normalized weight value to update the normalized weight value; updating the weight value of the perception data type using the updated normalized weight value; the preset condition includes a second preset time and / or the fluctuation of the historical accuracy of determining personnel location exceeding a preset fluctuation range. The calculation unit 703 is used to calculate the time decay factor of the sensing data corresponding to the sensing data type, and to calculate the address confidence level corresponding to the address pointed to in the sensing data. The first determining unit 704 is used to determine the comprehensive score of the address based on the weight value of the perceptual data type corresponding to the address, the time decay factor of the perceptual data of the perceptual data type corresponding to the address, and the address confidence. The second determining unit 705 is used to filter out the address with the highest comprehensive score based on the comprehensive score of all the addresses, and determine it as the location of key personnel.

[0174] It should be noted that other corresponding descriptions of the functional units involved in the key personnel location determination device provided in this application embodiment can be found by referring to... Figure 1 The corresponding description in [the document] will not be repeated here.

[0175] In an exemplary embodiment, a computer device is also provided, comprising a bus, a processor, a memory, and a communication interface. It may also include an input / output interface and a display device, wherein the various functional units can communicate with each other via the bus. The memory stores a computer program, and the processor executes the program stored in the memory to perform a method for determining the location of key personnel as described in the above embodiment.

[0176] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for determining the location of key personnel.

[0177] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented in hardware or by using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0178] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application.

[0179] Those skilled in the art will understand that the modules in the apparatus of the implementation scenario can be distributed within the apparatus of the implementation scenario as described, or they can be located in one or more apparatuses different from this implementation scenario, with corresponding changes. The modules of the above-described implementation scenario can be combined into one module, or they can be further divided into multiple sub-modules.

[0180] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of the implementation scenario.

[0181] The above disclosures are only a few specific implementation scenarios of this application. However, this application is not limited to these. Any variations that can be conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for determining the location of key personnel, characterized in that, include: Acquire dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types. Obtain the weight values ​​of the perceived data type; The step of determining the weight value of the perception data type includes: when a preset condition is met, determining the effective sample size of historical perception data for different perception data types; if the effective sample size of historical perception data for a perception data type is lower than a corresponding preset number, then setting the weight value of the perception data type to a preset static weight value; if the effective sample size of historical perception data for a perception data type is not lower than a corresponding preset number, then setting the weight value of the perception data type based on the historical accuracy of determining personnel location, historical average time decay factor, and average spatial correlation of the perception data type; normalizing the weight values ​​of all perception data types so that the sum of all normalized weight values ​​equals the total number of types of all perception data types; smoothly transitioning the normalized weight value to the previously obtained normalized weight value to update the normalized weight value; updating the weight value of the perception data type using the updated normalized weight value; the preset condition includes a second preset time and / or the fluctuation of the historical accuracy of determining personnel location exceeding a preset fluctuation range. Calculate the time decay factor of the sensed data, and calculate the address confidence level corresponding to the address pointed to by the sensed data; All addresses pointed to by the sensing data in the multi-source sensing data are sequentially taken as target addresses. Based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, the comprehensive score of the target address is determined. Based on the overall score of all the target addresses, the target address with the highest overall score is selected and identified as the location of key personnel.

2. The method according to claim 1, characterized in that, Before the step of setting the weight value of the perception data type based on the accuracy of determining personnel location, historical average time decay factor, and average spatial correlation degree according to the history of the perception data type, the method further includes: The accuracy rate of determining personnel location based on the number of times the historical data of the perception data type correctly determined the personnel location and the number of times the historical data of the perception data type incorrectly determined the personnel location is calculated. Based on the effective sample size and the historical time decay factor corresponding to all historical sensing data of the sensing data type, calculate the historical average time decay factor. The average spatial correlation degree is calculated based on the straight-line distance between the address and the center of the historical activity area of ​​the person in the historical sensing data of the aforementioned sensing data type.

3. The method according to claim 1, characterized in that, Before the step of sequentially taking all addresses pointed to by the sensing data in the multi-source sensing data as target addresses, and determining the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, the method further includes: Sensing data with a time decay factor less than a preset time decay factor and sensing data with an address confidence level less than a preset address confidence level are removed to update the multi-source sensing data.

4. The method according to claim 1, characterized in that, Also includes: Determine whether the distance between any two addresses pointed to by the multi-source sensing data exceeds a preset distance; A data conflict group is identified, which includes any two addresses to which the sensing data points are more than a preset range apart; the distance between sensing data in different data conflict groups does not exceed the preset range. Determine the overall score of the addresses in each of the data conflict groups; The addresses in each data conflict group are ranked from highest to lowest based on their overall scores, and the address with the highest overall score in each data conflict group is determined. Sort the addresses in all data conflict groups by their highest comprehensive scores from highest to lowest, filter out the top N highest comprehensive scores, and calculate the probability of the addresses corresponding to the top N highest comprehensive scores.

5. The method according to claim 1, characterized in that, The calculation of the time decay factor of the sensed data is performed according to the following formula: = ； in, The time decay factor is the kth sensing data corresponding to the sensing data type. The time difference is the difference between the time when the current location of the person is determined and the time when the kth sensing data is collected. This is the attenuation coefficient.

6. The method according to claim 1, characterized in that, The calculation of the address confidence level corresponding to the address pointed to by the sensing data is performed according to the following formula: ； in, For address; Address confidence; Address matching rate; Address standardization levels; To improve the accuracy of equipment data acquisition; These are the fusion weights for address matching rate, address standardization level, and device acquisition accuracy, respectively. .

7. The method according to claim 1, characterized in that, The comprehensive score of the target address is determined based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address, according to the following formula: ； in, For the target address; This is the overall score for the target address; The set of all perceptual data types that point to the target address; For the first Under a given perceptual data type, it is the collection of all perceptual data pointing to the target address; This is the weight value of the i-th type of sensing data corresponding to the k-th sensing data pointing to the target address; For the first Under the first perceptual data type, the first one pointing to the target address Time decay factor of sensor data; Address confidence level for the target address.

8. The method according to claim 1, characterized in that, Also includes: Determine whether the location of the key personnel exceeds a preset range; When the preset range is exceeded, the current location and basic information of the key personnel are pushed to the monitoring personnel; it is determined whether the location of the key personnel is outside the district / county boundary and is in an unreported state; if the location of the key personnel is outside the district / county boundary and is in an unreported state, the current location and basic information of the key personnel are pushed to the superior unit of the monitoring personnel. The locations of key personnel determined at different times are used to form spatiotemporal trajectories, which are then visualized. Based on the spatiotemporal trajectories of the key personnel, identify their companions.

9. A device for determining the location of key personnel, characterized in that, include: The first acquisition unit is used to acquire the dynamic information of key personnel within a first preset time period. The dynamic information of personnel includes multi-source sensing data, which includes sensing data of different types. The second acquisition unit is used to acquire the weight value of the perceived data type; The step of determining the weight value of the perception data type includes: when a preset condition is met, determining the effective sample size of historical perception data for different perception data types; if the effective sample size of historical perception data for a perception data type is lower than a corresponding preset number, then setting the weight value of the perception data type to a preset static weight value; if the effective sample size of historical perception data for a perception data type is not lower than a corresponding preset number, then setting the weight value of the perception data type based on the historical accuracy of determining personnel location, historical average time decay factor, and average spatial correlation of the perception data type; normalizing the weight values ​​of all perception data types so that the sum of all normalized weight values ​​equals the total number of types of all perception data types; smoothly transitioning the normalized weight value to the previously obtained normalized weight value to update the normalized weight value; updating the weight value of the perception data type using the updated normalized weight value; the preset condition includes a second preset time and / or the fluctuation of the historical accuracy of determining personnel location exceeding a preset fluctuation range. The calculation unit is used to calculate the time decay factor of the sensed data and the address confidence level corresponding to the address pointed to by the sensed data. The first determining unit is used to sequentially take all addresses pointed to by the sensing data in the multi-source sensing data as target addresses, and determine the comprehensive score of the target address based on the weight value of the sensing data type corresponding to the sensing data pointing to the target address, the time decay factor of the sensing data pointing to the target address, and the address confidence of the target address. The second determining unit is used to filter out the target address with the highest comprehensive score based on the comprehensive score of all the target addresses, and determine it as the location of key personnel.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for determining the location of key personnel as described in any one of claims 1 to 8.

Images