A method for verifying patrol personnel based on multi-node perception and collaboration
By deploying multi-node sensing devices in the patrol area, and collaboratively collecting and analyzing multi-source data from patrol personnel, the problem of the inability to verify the authenticity of patrols in the existing system has been solved, and highly accurate verification of patrol behavior has been achieved.
Patent Information
- Application Number
- CN202611421940.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-14
- Publication Date
- 2026-11-10
AI Technical Summary
The existing intelligent patrol system lacks a joint analysis mechanism for the consistency of time, space and identity information among different sensing nodes, which makes it impossible to effectively identify proxy clocking and fake patrols.
By deploying various types of sensing nodes in the patrol area, the system collaboratively collects the identity information, location change information, and behavioral status information of patrol personnel. Based on time consistency, spatial consistency, and identity consistency, it establishes multi-node association relationships to form a multi-node evidence chain to verify the authenticity of the patrol.
It effectively identifies proxy clocking and fake patrols, improving the accuracy and reliability of verifying the authenticity of patrol personnel's identity, location, and behavior.
Smart Images

Figure CN123072107A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of personnel behavior verification technology, and more specifically, to a method for verifying patrol personnel based on multi-node perception and collaboration. Background Technology
[0002] With the development of smart community and intelligent property management technologies, residential property management companies typically need to arrange security personnel to patrol public areas, underground parking garages, equipment rooms, and key security areas according to prescribed times and routes to promptly identify security risks and ensure the safety of residents' living environment. To improve patrol management efficiency, electronic patrol systems are currently commonly used to record the patrol process. Existing intelligent patrol systems typically deploy RFID tags, NFC tags, QR code markers, Bluetooth positioning nodes, video surveillance equipment, or access control devices within the patrol area. Patrol personnel carry mobile terminals, electronic tag readers, or identity recognition devices to collect identity information, location data, and time information during the patrol process. The system uploads the collected data to a backend server and manages the completion status of patrol tasks by determining whether patrol personnel have reached designated patrol points, met prescribed time requirements, and completed the patrol according to the preset route.
[0003] For example, Chinese patent CN107633567A discloses an intelligent patrol method based on electronic tags and mobile terminals, such as... Figure 2 As shown, this solution records the patrol process by setting up multiple patrol markers and using mobile terminals to collect patrol point information. For example, existing personnel patrol technology solutions based on positioning and video analysis determine personnel location and activity status by fusing positioning information and video detection results. However, while existing intelligent patrol systems can collect the location, identity, and time information of patrol personnel using different types of sensing devices, each sensing node typically completes data collection independently, judging patrol status only based on a single patrol point or a single positioning result. There is a lack of a joint analysis mechanism to address the consistency of time, space, and identity information among different sensing nodes. This makes it impossible to effectively determine whether multiple sensing results correspond to the same genuine patrol behavior, leading to problems such as proxy clocking, fake patrols, and unverifiable patrol processes. Summary of the Invention
[0004] To overcome the aforementioned deficiencies in the prior art, this invention provides a method for verifying patrol personnel based on multi-node sensing collaboration. This method involves deploying various types of sensing nodes within the patrol area to collaboratively collect the identity information, location change information, and behavioral status information of patrol personnel during their patrols. Based on temporal consistency, spatial consistency, and identity consistency, data associations are established between different sensing nodes to form a multi-node evidence chain that can prove the authenticity of the patrol, thereby solving the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] A method for verifying patrol personnel based on multi-node sensing and collaboration includes the following steps:
[0007] Establish a patrol task model based on the patrol task requirements, and construct a multi-node perception network based on the target patrol area.
[0008] Multi-source sensing data is collected by using multiple types of sensing nodes during the patrol personnel's patrol tasks.
[0009] Based on temporal consistency, spatial consistency, and identity consistency, multi-source perception data are correlated layer by layer to establish multi-node association relationships.
[0010] By integrating the identity, time, space, and behavioral information of patrol personnel based on the multi-node association results, a multi-node evidence chain corresponding to the same patrol behavior is generated.
[0011] The patrol credibility of a patrol event is calculated based on node integrity, temporal consistency, spatial consistency, identity consistency, and behavioral integrity.
[0012] The patrol credibility is compared with the preset verification threshold, and the corresponding patrol verification result is output by combining the multi-node consistency results.
[0013] As a further aspect of the present invention, the step of establishing multi-node association relationships by progressively associating multi-source sensing data based on time consistency, spatial consistency, and identity consistency includes: progressively associating multi-source sensing data collected by multiple sensing nodes during the patrol process; performing time consistency analysis based on the generation time of data corresponding to different sensing nodes; and classifying multi-source sensing data that meet the time association conditions into the same candidate patrol event set; based on the candidate patrol event set, performing spatial consistency analysis according to the installation location, coverage area, and corresponding patrol area of each sensing node to determine whether different sensing results correspond to the same patrol area or the same continuous patrol process; and, based on satisfying time consistency and spatial consistency, performing identity consistency analysis on the identity information collected by different sensing nodes to determine whether multiple sensing results correspond to the same patrol personnel, and associating and saving the analysis results of time consistency, spatial consistency, and identity consistency to form a multi-node association result.
[0014] As a further aspect of the present invention, the time consistency analysis includes establishing a time correlation window between different sensing nodes based on the data generation characteristics of each sensing node and the movement process of patrol personnel in different patrol areas. The time correlation window is used to characterize the maximum allowable time deviation between the corresponding data of different sensing nodes. When determining the time correlation window, at least the acquisition delay of sensing devices, the data transmission delay of sensing data, the spatial distance between different sensing nodes, and the movement speed of patrol personnel between adjacent patrol areas are considered. When the time deviation between the corresponding data of different sensing nodes does not exceed the corresponding time correlation window, it is determined that the data corresponding to the different sensing nodes meets the time consistency condition.
[0015] As a further aspect of the present invention, the time association window is determined based on the maximum data response difference between different sensing nodes and the normal movement time of patrol personnel between adjacent sensing areas. For the facial recognition device and RFID electronic tag reading device at the entrance of the underground parking garage, the time association window is set to 10 seconds; for the facial recognition device at the entrance of the underground parking garage and the video surveillance device inside the underground parking garage, the time association window is set to 60 seconds. The generation time of the data corresponding to the i-th sensing node and the j-th sensing node is obtained, and the time deviation between the two is determined. When the time deviation is not greater than the time association window corresponding to the i-th sensing node and the j-th sensing node, it is determined that the data corresponding to the two sensing nodes meet the time consistency condition.
[0016] As a further aspect of the present invention, the patrol task model includes the identity information of the target patrol personnel, patrol time, patrol area, and patrol route. The patrol route is used to represent the order of patrols among multiple patrol areas. The identity perception node includes a facial recognition camera, an RFID electronic tag reader, and a mobile terminal carried by the patrol personnel. The location perception node includes a Bluetooth positioning node and an area entry detection device. The behavior perception node includes a video surveillance device set up within the patrol area.
[0017] As a further aspect of the present invention, the spatial consistency analysis includes establishing a spatial mapping relationship between sensing nodes and associating and saving the installation location, coverage area, and corresponding patrol area of each sensing node.
[0018] As a further aspect of the present invention, the identity consistency analysis includes pre-establishing an identity association relationship between patrol personnel identity information, facial identity identifier, RFID electronic tag number and mobile terminal identity identifier, and matching the facial identity identifier output by the facial recognition node, the RFID electronic tag number read by the RFID electronic tag reading device and the mobile terminal identity identifier with the identity association relationship respectively.
[0019] As a further aspect of the present invention, the multi-node evidence chain includes an identity association chain, a time association chain, a spatial association chain, and a behavior association chain.
[0020] As a further aspect of the present invention, the identity association chain records the face recognition result, RFID electronic tag information and mobile terminal identity identifier; the time association chain records the identity recognition time, tag reading time, positioning time and behavior detection time; the spatial association chain records the entrance area, patrol area and movement path passed by the patrol personnel; and the behavior association chain records the stay time, movement trajectory and patrol behavior of the patrol personnel after entering the patrol area.
[0021] The technical effects and advantages of this invention, a patrol personnel verification method based on multi-node perception and collaboration, are as follows: This invention collects patrol data collaboratively through identity perception nodes, location perception nodes, and behavior perception nodes, and establishes multi-node associations based on time consistency, spatial consistency, and identity consistency to form a multi-node evidence chain corresponding to the same patrol behavior. Then, the patrol credibility is calculated by combining node completeness, behavior completeness, etc., thereby effectively identifying cases of proxy clocking, fake patrols, and incomplete patrols, and improving the accuracy and reliability of verifying the authenticity of patrol personnel's identity, location, and behavior. Attached Figure Description
[0022] Figure 1 This is a flowchart of a patrol personnel verification method based on multi-node perception and collaboration according to the present invention.
[0023] Figure 2 This is a flowchart of the existing intelligent patrol method.
[0024] Figure 3 This is a schematic diagram illustrating the relationship between the patrol task model and the multi-node perception network construction of this invention.
[0025] Figure 4 This is a schematic diagram of the spatial mapping and spatial consistency determination of the sensing nodes in this invention.
[0026] Figure 5 This is a schematic diagram illustrating the association of multi-node identity information of patrol personnel according to the present invention.
[0027] Figure 6 This is a schematic diagram of the patrol task model and multi-node perception network simulation interface of the present invention.
[0028] Figure 7 This is a schematic diagram of the node distribution and simulation scene selection interface for the patrol area of this invention.
[0029] Figure 8 This is a schematic diagram of the simulation interface for multi-source sensing data acquisition and multi-node consistency analysis of the present invention.
[0030] Figure 9 This is a schematic diagram of the simulation interface for calculating the reliability of patrols and verifying patrol results in this invention. Detailed Implementation
[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0032] Example 1
[0033] See Figure 1 This invention provides a method for verifying patrol personnel based on multi-node perception and collaboration, which includes the following steps:
[0034] Step S1: Establish a patrol task model according to the patrol task requirements, and build a multi-node perception network based on the target patrol area.
[0035] Step S2: Use multiple types of sensing nodes to collect multi-source sensing data during the patrol personnel's patrol tasks.
[0036] Step S3: Based on time consistency, spatial consistency and identity consistency, multi-source perception data are correlated layer by layer to establish multi-node association relationships.
[0037] Step S4: Based on the multi-node association results, integrate the identity, time, space and behavior information of the patrol personnel to generate a multi-node evidence chain corresponding to the same patrol behavior.
[0038] Step S5: Calculate the patrol credibility corresponding to the patrol event based on node integrity, time consistency, spatial consistency, identity consistency, and behavior integrity.
[0039] Step S6: Compare the patrol credibility with the preset verification threshold, and output the corresponding patrol verification result based on the multi-node consistency result.
[0040] Further, in step S1, a patrol task model is established according to the patrol task requirements, and a multi-node sensing network is constructed based on the target patrol area. In this embodiment, a patrol task model is first established based on the actual regional structure of the target residential community and the patrol requirements formulated by the property management department to clarify the target patrol personnel, patrol time, patrol area, and patrol route corresponding to this patrol task. The above information is used as the benchmark for subsequent multi-node sensing data association and patrol authenticity verification. In this embodiment, the residential community includes underground parking areas, building public areas, power distribution equipment areas, and fire lane areas. The property management department arranges security personnel A to perform night patrol tasks, and sets 22:00 to 23:00 as the effective execution time for this patrol task. The underground garage, fire lane, and equipment room are identified as the target patrol areas that need to be verified during this patrol process. At the same time, a preset patrol route is formed in the order of underground garage, fire lane, and equipment room. Therefore, the identity information of security personnel A, the patrol time range from 22:00 to 23:00, the patrol area information corresponding to the underground garage, fire exits, and equipment rooms, and the order of these patrol areas are associated and saved to form a patrol task model corresponding to this night patrol mission. When subsequently acquiring patrol data generated by each sensing node, this patrol task model is used as the matching basis to determine whether the collected data meets the requirements of this patrol mission based on aspects such as personnel identity, patrol time, personnel location, and movement order.
[0041] After establishing the patrol task model, based on the patrol route and the location of each target patrol area determined in the patrol task model, sensing nodes are configured in the underground garage, fire exits, and equipment rooms to collect information from patrol personnel in different dimensions, thereby constructing a multi-node sensing network. In this embodiment, the sensing nodes include identity sensing nodes, location sensing nodes, and behavior sensing nodes. The identity sensing nodes utilize facial recognition cameras, RFID tag readers, and mobile terminals carried by patrol personnel. The facial recognition cameras are positioned at the entrance of the target patrol area or at locations where personnel normally pass through, to obtain facial recognition results of personnel entering the corresponding area. The RFID tag readers are positioned at personnel passage locations within the corresponding patrol area to read the RFID tags carried by patrol personnel. The mobile terminals are pre-bound to the identity of security personnel A, providing a mobile terminal identity identifier corresponding to this patrol task.
[0042] Location-aware nodes utilize Bluetooth positioning nodes and area entry detection devices deployed in underground parking garages, fire exits, and equipment rooms to obtain information on the time, location, and positional changes of patrol personnel entering different patrol areas. Behavior-aware nodes employ video surveillance equipment located within these patrol areas to capture the stationary and movement status of patrol personnel after entering the target area, as well as the patrol process itself. This setup enables the generation of identity-aware, location-aware, and behavior-aware data when the same patrol personnel passes through a target patrol area, providing a data foundation for subsequently determining whether different perception results correspond to the same actual patrol behavior. See also... Figure 3 This illustrates the construction relationship between the patrol task model and various types of sensing nodes in each target patrol area.
[0043] Further, in step S2, multi-source sensing data is collected during the patrol personnel's patrol task using multiple types of sensing nodes. Specifically, after security personnel A begins to perform the patrol task within the time range specified by the patrol task model, each sensing node in the multi-node sensing network acquires multi-source sensing data during the patrol process according to its own data acquisition method. This embodiment takes the process of security personnel A entering the underground parking garage for patrol as an example. At 22:05:10, the facial recognition camera at the entrance of the underground parking garage detects the person entering and obtains the facial recognition result corresponding to security personnel A; at 22:05:12, the RFID electronic tag reader at the personnel passage location in the underground parking garage reads the electronic tag A001 carried by security personnel A; at 22:05:15, the Bluetooth positioning node in the underground parking garage detects that the mobile terminal bound to security personnel A has entered the positioning area corresponding to the underground parking garage; after 22:05:20, the video surveillance equipment continuously acquires the movement and dwelling status of security personnel A inside the underground parking garage and records their movement along the patrol path and their patrol behavior within that area. For the aforementioned data, the corresponding sensing node type, data generation time, corresponding spatial area, and identifiable personnel identity or behavioral status are recorded to form a multi-source sensing data set corresponding to this patrol process in the underground parking garage. Since the above data are generated by different types of devices and the data collection times are not exactly the same, all data cannot be directly identified as belonging to the same patrol behavior. Instead, it is necessary to continue to correlate them layer by layer through time consistency, spatial consistency, and identity consistency.
[0044] Further, in step S3, multi-source sensing data is correlated layer by layer based on time consistency, spatial consistency, and identity consistency to establish multi-node association relationships. After obtaining multi-source sensing data during the patrol process, due to differences in data acquisition methods, data transmission processes, and data processing times of different types of sensing devices, even if multiple sensing nodes correspond to the same actual patrol behavior, the timestamps of the generated data may have certain deviations. Therefore, this embodiment first analyzes the time relationship of data collected by different sensing nodes to establish a preliminary association relationship based on time consistency. Specifically, based on the data generation characteristics of each sensing node and the normal movement process of patrol personnel in different patrol areas, a time association window determination model is pre-established to determine the maximum allowable time deviation between the data of different sensing nodes. The time association window determination model considers at least the following factors: sensing device acquisition delay; sensing data transmission delay; spatial distance between different sensing nodes; and normal movement speed of patrol personnel between adjacent patrol areas. For different types of sensing nodes, due to their different data generation methods, the corresponding data response times differ. For example, facial recognition cameras typically generate identification data immediately upon a person entering the monitored area, RFID tag readers generate tag data when the tag enters the reading range, while Bluetooth positioning nodes need to obtain location data based on the terminal's broadcast cycle. Therefore, directly using fixed time points for data matching can easily lead to data belonging to the same patrol activity being unable to be correlated.
[0045] This embodiment determines the time correlation window based on the maximum data response difference between different sensing nodes and the normal movement time of patrol personnel between adjacent sensing areas. Specifically, the time when the i-th sensing node generates data is set to... The time when the j-th sensing node generates data is The time deviation between the two sensing nodes is: Δ =| - |, when the time deviation Δ Satisfy: Δ ≤ If the data generated by the two sensing nodes satisfies the time consistency condition, then... This represents the time association window between the i-th and j-th sensing nodes, which is dynamically determined based on the data response delay and spatial distance between the two sensing nodes. For example: If the facial recognition device and RFID reader at the entrance of the underground parking garage are close together, and the time interval between a person passing through the two devices is short, then the corresponding time association window is set to 10 seconds. If the facial recognition device at the entrance of the underground parking garage is far from the video surveillance equipment inside the garage, and it takes time for a person to move from the entrance to the equipment, then the corresponding time association window is set to 60 seconds. In this embodiment, when security personnel A enters the underground parking garage, the facial recognition device detects the person's identity information at 22:05:10, the RFID reader reads the electronic tag A001 at 22:05:12, and the Bluetooth positioning node detects the bound mobile terminal entering the underground parking garage at 22:05:15. Specifically: the time difference between facial recognition data and RFID data is |22:05:12-22:05:10| = 2 seconds; the time difference between facial recognition data and location data is |22:05:15-22:05:10| = 5 seconds. Since these time deviations are all less than the time association window between the corresponding sensing nodes, the data is determined to meet the time consistency condition, and the data is classified into the same candidate patrol event set. This time consistency analysis method avoids incorrect association of data generated at different time stages and can adapt to data delays between different sensing devices, improving the accuracy of multi-node association.
[0046] After completing the time consistency analysis and obtaining candidate association data, spatial consistency judgment is further performed based on the spatial location relationships of different sensing nodes to determine whether different sensing results correspond to the same patrol area or the same continuous patrol process. Since different sensing nodes have different detection ranges—for example, facial recognition devices typically correspond to fixed entrance areas; RFID reading devices correspond to tag reading ranges; Bluetooth positioning nodes correspond to certain spatial coverage areas; and video surveillance devices correspond to video acquisition areas—this embodiment first establishes a spatial mapping model for sensing nodes, associating and saving the installation location, coverage range, and corresponding patrol area of each sensing node. Specifically, for any sensing node Nk, a corresponding spatial region is established: Sk = {location coordinates, coverage radius, patrol area}. Here, the location coordinates represent the installation location of the sensing node; the coverage radius represents the range within which the node can effectively collect data; and the patrol area represents the location of the patrol task corresponding to the node. Multiple sensing nodes are considered to have spatial consistency when the spatial areas corresponding to them meet any of the following conditions: (1) the coverage areas of multiple sensing nodes overlap; (2) the patrol areas corresponding to multiple sensing nodes belong to the same target patrol area; (3) there is a preset patrol route connection between the areas corresponding to multiple sensing nodes. Furthermore, when the location sensing node can output the real-time location of the patrol personnel, spatial matching is performed based on the distance between the personnel's location and the coverage areas of different nodes. Let the current location of the patrol personnel be... The sensing node covers the central location as Then calculate the spatial distance between the current location of the patrol personnel and the coverage center of the sensing node: ,when When R is within the effective detection range of the sensing node, the patrol personnel are considered to be within that node's effective coverage distance. In this embodiment, the facial recognition device at the underground parking garage entrance covers the underground parking garage entrance area; the RFID reader also covers the underground parking garage entrance area; and the Bluetooth positioning node covers the interior area of the underground parking garage. When: the facial recognition device detects security personnel A entering the underground parking garage; the RFID reader reads the corresponding electronic tag; the Bluetooth positioning node detects that security personnel A's bound terminal is within the underground parking garage area; and the video surveillance device detects personnel moving from the entrance into the parking garage. Since the spatial areas corresponding to the above multiple sensing nodes satisfy a continuous connection relationship, the above sensing data is determined to have spatial consistency. Through spatial consistency analysis, data misassociations caused by similar identity information or identical tag information in different areas can be avoided, improving the ability to judge the spatial authenticity of the patrol process. See also Figure 4 This further illustrates the spatial mapping relationship between each sensing node and the target patrol area, as well as the process for determining spatial consistency among different sensing nodes.
[0047] Based on satisfying temporal and spatial consistency, further identity consistency analysis is performed on the data collected by different sensing nodes to determine whether multiple sensing results belong to the same patrol personnel. In this embodiment, a patrol personnel identity association table is pre-established, binding patrol personnel identity information, facial recognition, RFID electronic tag number, and mobile terminal identity. For example, Table 1 shows the patrol personnel identity association table:
[0048] Table 1. Correspondence between patrol personnel identity information and multi-node identity identifiers
[0049]
[0050] When performing identity consistency analysis: First, the facial identity identifier output by the facial recognition node is matched with the patrol personnel identity database; second, the electronic tag number read by the RFID node is matched with the identity association table; third, the mobile terminal identity identifier is compared with the above identity information. If: Facial identity = RFID-bound identity = Mobile terminal identity, then identity consistency is confirmed.
[0051] In this embodiment, the facial recognition result is: Security Personnel A; the RFID reading result is: A001; and the mobile terminal identity is: Terminal_A. After identity association, all three identity information corresponds to Security Personnel A, thus determining that this group of data belongs to the same patrol personnel. If the following occurs: the RFID tag belongs to Security Personnel A; but the facial recognition result belongs to Security Personnel B; or the mobile terminal location is inconsistent with the tag location; then the identity consistency is considered invalid, and the credibility of the patrol event is reduced. Through identity consistency analysis, the problem of proxy clocking caused by relying solely on electronic tags or mobile terminals for patrol confirmation can be effectively avoided. After completing the time consistency, spatial consistency, and identity consistency analysis, the data generated by each sensing node and the corresponding consistency analysis results are associated and saved using the candidate patrol event set as a unit, forming a multi-node association result. The multi-node association result includes the established consistency association relationships between different sensing nodes, as well as the inconsistent states formed due to missing sensing data, incompatibility, or conflicts between different sensing results. Thus, it is possible to uniformly form evidence records for normal associated data, missing data, and conflicting data in the same candidate patrol event. See also Figure 5 This illustrates the association of multi-node identity information for patrol personnel according to the present invention.
[0052] Further, in step S4, the identity, time, space, and behavior information of patrol personnel are integrated based on the multi-node association results to generate a multi-node evidence chain corresponding to the same patrol behavior. Specifically, taking the candidate patrol event set formed in step S3 as a unit, the identity perception data, location perception data, and behavior perception data generated by each sensing node, as well as the corresponding time consistency, space consistency, and identity consistency analysis results, are associated and saved to form a multi-node evidence chain corresponding to the candidate patrol event. For sensing data that meets the corresponding consistency conditions, the corresponding association relationship is recorded in the multi-node evidence chain; for cases where sensing data is missing, the corresponding consistency relationship cannot be established, or there are conflicts between data from different sensing nodes, the corresponding data missing state or inconsistency state is recorded in the multi-node evidence chain as the basis for subsequent patrol credibility calculation and patrol verification result determination. In this embodiment, the multi-node evidence chain includes:
[0053] 1. Identity association chain, used to record the identity information of personnel performing patrol tasks, including: facial recognition results; RFID tag information; mobile terminal identity.
[0054] 2. The time-related chain is used to record the correspondence between the time of data generation of different sensing nodes, including: identity recognition time; tag reading time; location time; and behavior detection time.
[0055] 3. Spatial association chain, used to record the areas actually traversed by patrol personnel, including: entrance area; patrol area; movement path.
[0056] 4. Behavioral association chain, used to record whether patrol personnel have performed actual patrol behaviors, including: entering the area; dwell time; movement trajectory; on-site operations.
[0057] In this embodiment, the following chain of evidence is formed: Security Personnel A → Identity verification at 22:05:10 → Tag reading at 22:05:12 → Entry into the underground parking garage at 22:05:15 → Patrol from 22:05:20 to 22:06:00. The above evidence collectively proves that Security Personnel A entered the underground parking garage at 22:05 and completed the actual patrol.
[0058] Further, in step S5, the patrol credibility of the patrol event is calculated based on node completeness, temporal consistency, spatial consistency, identity consistency, and behavioral completeness. After forming a multi-node evidence chain corresponding to the current patrol event, the current patrol event is comprehensively scored based on the data acquisition status of various sensing nodes in the multi-node evidence chain, as well as the analysis results of temporal consistency, spatial consistency, and identity consistency obtained in step S3, and the completeness of behavioral sensing data. For cases where sensing node data is missing or the corresponding consistency relationship is not established, the corresponding score is reduced according to the corresponding scoring item, thereby determining the patrol credibility of the current patrol event. This embodiment uses a comprehensive scoring method to calculate the patrol credibility, with a total score of 100 points. Among them: 1. Node completeness score. Node completeness is used to evaluate whether the different types of sensing nodes participating in the formation of the multi-node evidence chain are complete. This embodiment evaluates according to three types: identity sensing nodes, location sensing nodes, and behavioral sensing nodes, with identity sensing nodes corresponding to 10 points, location sensing nodes corresponding to 10 points, and behavioral sensing nodes corresponding to 10 points. When all three types of sensing nodes—identity-aware nodes, location-aware nodes, and behavior-aware nodes—have corresponding sensing data and participate in the multi-node association of the current patrol event, the node completeness score is 30 points. If any type of sensing node data is missing, 10 points are deducted from the score for that type. For example, if RFID tag reading data and location data exist in the current patrol event, but no corresponding behavior-aware data is obtained, the behavior-aware node type is determined to be missing, and 10 points are deducted from the node completeness score.
[0059] 2. Time consistency score, with a maximum score of 20 points. Determined based on time deviation: Δ ≤20% , get 20 points; when 20% , get 10 points; when 0 points.
[0060] 3. Spatial consistency score, with a maximum score of 20 points. The following scores apply: 20 points if multiple nodes correspond to the same patrol area; 15 points if multiple nodes correspond to locations that do not belong to the same patrol area but have a continuous path relationship along a preset patrol route; 10 points if multiple nodes correspond to spatially adjacent areas but cannot form a complete continuous path relationship; and 0 points if multiple nodes correspond to locations that do not belong to the same area, have a continuous path, or have a spatially adjacent relationship.
[0061] The spatial consistency is determined based on the deployment location, coverage area, patrol area, and patrol personnel location obtained by each sensing node.
[0062] 4. Identity consistency score, with a maximum score of 20 points. When: facial recognition information, RFID electronic tag information, and mobile terminal identity information all correspond to the same patrol personnel: 20 points; when two types of identity information correspond to the same patrol personnel, and the third type is missing: 10 points; when identity information from different nodes conflicts or cannot be matched to the same patrol personnel: 0 points.
[0063] Among them, the identity consistency score is used to evaluate whether multiple independent identity sources jointly point to the target patrol personnel specified in the patrol task model.
[0064] 5. Behavioral Completeness Score, with a maximum score of 10 points. A score of 10 points is awarded when video detection simultaneously meets the following criteria: entry into the target patrol area; dwell time meeting the patrol requirements for the corresponding patrol area; and continuous movement or patrolling behavior within the target patrol area. A score of 5 points is awarded if only patrol personnel are detected entering the target patrol area, but no dwelling or patrolling behavior meeting the patrol requirements is observed. A score of 0 points is awarded if no personnel behavior corresponding to the current patrol event is detected.
[0065] Finally: C = C1 + C2 + C3 + C4 + C5, where: C is the patrol credibility; C1 is the node completeness score; C2 is the temporal consistency score; C3 is the spatial consistency score; C4 is the identity consistency score; and C5 is the behavioral completeness score. In this embodiment: Security personnel A's patrol event: Node completeness: 30 points; Temporal consistency: 20 points; Spatial consistency: 20 points; Identity consistency: 20 points; Behavioral completeness: 10 points. The overall result is: Patrol credibility: 100 points.
[0066] Further, in step S6, the patrol credibility is compared with a preset verification threshold, and the corresponding patrol verification result is output based on the multi-node consistency result. After obtaining the patrol credibility corresponding to the current patrol event, the patrol credibility is compared with a preset effective patrol threshold and anomaly judgment threshold, and the corresponding patrol verification result is output based on whether time consistency, spatial consistency, and identity consistency are satisfied. In this embodiment, the effective patrol threshold is set to 80 points, and the anomaly judgment threshold is set to 60 points. When the patrol credibility is not lower than 80 points, time consistency, spatial consistency, and identity consistency are all satisfied, and the behavior integrity score reaches 10 points, the current patrol event is judged as a valid patrol, indicating that the patrol personnel specified by the patrol task model actually arrived at the corresponding patrol area within the specified time, and there is a patrol behavior that can be associated with the personnel's identity. Taking the aforementioned underground parking garage patrol incident as an example, when the facial identity of security personnel A matches the RFID electronic tag A001, the location result indicates that they have entered the underground parking garage, the video surveillance equipment detects that they stay and patrol in the area, and when the final patrol credibility reaches 100 points, a valid patrol result is output.
[0067] When the patrol credibility score is not lower than 60 points but lower than 80 points, or when the patrol credibility score is not lower than 80 points but the behavior completeness does not meet the preset behavior verification conditions, or when there are missing perception data or insufficient consistency relationships in the multi-node evidence chain, the current patrol event is judged as a suspected abnormal patrol, and the corresponding missing data, inconsistent relationships, or unmet behavior verification conditions are used as the basis for abnormal prompts. For example, if the RFID electronic tag reader normally reads security personnel A's A001, but no facial recognition result of security personnel A is detected in the same time association window and corresponding spatial area, or the video surveillance equipment does not obtain behavior data that can prove that the person actually entered and patrolled, although some nodes can indicate that the relevant identity certificate has been to the patrol area, it is not possible to fully prove that security personnel A himself performed the patrol through the multi-node evidence chain. Therefore, a suspected abnormal patrol result is output to indicate that there is a possibility that the electronic tag was carried by other people or that someone else clocked in on his behalf.
[0068] When the patrol credibility score is below 60 points, or when both identity consistency and spatial consistency fail to hold, making it impossible for multiple sensing results to form a valid correlation proving that the target patrol personnel actually arrived at the target patrol area, the current patrol event is judged as an incomplete patrol. For example, although the RFID electronic tag reader reads A001, the facial recognition device does not detect security personnel A within the same time range, and Bluetooth positioning data indicates that the mobile terminal bound to security personnel A has not entered the underground parking garage area. Therefore, the RFID electronic tag data cannot form a complete temporal, spatial, and identity correlation with the personnel identity data and location data. The corresponding multi-node evidence chain cannot prove that security personnel A actually arrived at the underground parking garage and performed a patrol. Therefore, the patrol event is judged as an incomplete patrol, and the corresponding patrol verification result is output. Through the above-mentioned hierarchical verification method, the system can not only determine whether a single check-in record exists, but also distinguish between valid patrols, suspected abnormal patrols, and incomplete patrols based on the joint correlation between different sensing nodes, thereby achieving a comprehensive verification of the authenticity of the patrol personnel's identity, location, and patrol behavior.
[0069] See Figures 6-9 To further illustrate the execution process of the above-mentioned multi-node perception collaborative patrol personnel verification method in a real software environment, this embodiment constructs a multi-node perception collaborative patrol personnel verification simulation platform based on the aforementioned patrol task model, multi-source perception data acquisition, multi-node association, multi-node evidence chain generation, patrol credibility calculation, and patrol verification result output process. This platform is used to dynamically simulate and visualize the execution process of the technical solution of this invention.
[0070] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
[0071] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for verifying patrol personnel based on multi-node sensing and collaboration, characterized in that, The method includes: Establish a patrol task model based on the patrol task requirements, and construct a multi-node perception network based on the target patrol area; Multi-source sensing data is collected by using multiple types of sensing nodes during the patrol personnel's patrol tasks. Based on temporal consistency, spatial consistency, and identity consistency, multi-source sensing data are correlated layer by layer to establish multi-node association relationships; By integrating the identity, time, space and behavior information of patrol personnel based on the multi-node association results, a multi-node evidence chain corresponding to the same patrol behavior is generated; The patrol credibility of a patrol event is calculated based on node integrity, temporal consistency, spatial consistency, identity consistency, and behavioral integrity. The patrol credibility is compared with the preset verification threshold, and the corresponding patrol verification result is output by combining the multi-node consistency results.
2. The method according to claim 1, characterized in that, The step-by-step association of multi-source sensing data based on time consistency, spatial consistency, and identity consistency to establish multi-node association relationships includes: performing layer-by-layer association of multi-source sensing data collected by multiple sensing nodes during patrols; conducting time consistency analysis based on the generation time of data corresponding to different sensing nodes; and grouping multi-source sensing data that meet the time association conditions into the same candidate patrol event set; based on the candidate patrol event set, conducting spatial consistency analysis according to the installation location, coverage area, and corresponding patrol area of each sensing node to determine whether different sensing results correspond to the same patrol area or the same continuous patrol process; and, based on satisfying time consistency and spatial consistency, conducting identity consistency analysis on the identity information collected by different sensing nodes to determine whether multiple sensing results correspond to the same patrol personnel, and associating and saving the analysis results of time consistency, spatial consistency, and identity consistency to form a multi-node association result.
3. The method according to claim 2, characterized in that, The time consistency analysis includes establishing a time correlation window between different sensing nodes based on the data generation characteristics of each sensing node and the movement process of patrol personnel in different patrol areas. The time correlation window is used to characterize the maximum allowable time deviation between the corresponding data of different sensing nodes. When determining the time correlation window, at least the acquisition delay of sensing devices, the data transmission delay of sensing data, the spatial distance between different sensing nodes, and the movement speed of patrol personnel between adjacent patrol areas are considered. When the time deviation between the corresponding data of different sensing nodes does not exceed the corresponding time correlation window, it is determined that the data corresponding to the different sensing nodes meets the time consistency condition.
4. The method according to claim 3, characterized in that, The time association window is determined based on the maximum data response difference between different sensing nodes and the normal movement time of patrol personnel between adjacent sensing areas. For the facial recognition device and RFID electronic tag reading device at the entrance of the underground parking garage, the time association window is set to 10 seconds. For the facial recognition device at the entrance of the underground parking garage and the video surveillance device inside the underground parking garage, the time association window is set to 60 seconds. The generation time of the data corresponding to the i-th sensing node and the j-th sensing node is obtained, and the time deviation between the two is determined. When the time deviation is not greater than the time association window corresponding to the i-th sensing node and the j-th sensing node, it is determined that the data corresponding to the two sensing nodes meet the time consistency condition.
5. The method according to claim 1, characterized in that, The patrol task model includes the identity information of the target patrol personnel, patrol time, patrol area, and patrol route. The patrol route is used to represent the order of patrols among multiple patrol areas. The multi-type sensing nodes include identity sensing nodes, location sensing nodes, and behavior sensing nodes. The identity sensing nodes include facial recognition camera equipment, RFID electronic tag reading equipment, and mobile terminals carried by patrol personnel. The location sensing nodes include Bluetooth positioning nodes and area entry detection equipment. The behavior sensing nodes include video surveillance equipment set in the patrol area.
6. The method according to claim 2, characterized in that, The spatial consistency analysis includes establishing a spatial mapping relationship between sensing nodes and associating and saving the installation location, coverage area, and corresponding patrol area of each sensing node.
7. The method according to claim 2, characterized in that, The identity consistency analysis includes pre-establishing an identity association relationship between patrol personnel's identity information, facial identity identifier, RFID electronic tag number, and mobile terminal identity identifier, and matching the facial identity identifier output by the facial recognition node, the RFID electronic tag number read by the RFID electronic tag reading device, and the mobile terminal identity identifier with the identity association relationship respectively.
8. The method according to claim 1, characterized in that, The multi-node evidence chain includes an identity association chain, a time association chain, a spatial association chain, and a behavior association chain.
9. The method according to claim 8, characterized in that, The identity association chain records the facial recognition result, RFID electronic tag information and mobile terminal identity identifier; the time association chain records the identity recognition time, tag reading time, positioning time and behavior detection time; the spatial association chain records the entrance area, patrol area and movement path passed by the patrol personnel; and the behavior association chain records the stay time, movement trajectory and patrol behavior of the patrol personnel after entering the patrol area.
Citation Information
Patent Citations
Intelligent night patrol method
CN107633567A
