Passive tag assisted limited space key equipment unmanned aerial vehicle inspection method and system

Abstract

The invention relates to a passive tag assisted limited space key equipment unmanned aerial vehicle inspection method and system. The system comprises the following steps: constructing an RFID tag database; the target limited space is explored, a three-dimensional environment model is constructed, and a label installation position is determined; installing an RFID tag in the target limited space; and based on the target limited space after the tag is installed, the unmanned aerial vehicle performs routing inspection by adopting an RFID identification judgment algorithm based on the reading times and RSSI double thresholds, and analyzes the routing inspection result. According to the invention, the passive RFID tag is used as a space anchor point and a task information source, low-cost and high-reliability autonomous navigation and equipment inspection are realized in a limited space environment with unstable GNSS signals, changeable illumination conditions and a complex structure, the inspection path is dynamically adjusted, the inspection operation is accurately triggered, and equipment position checking and responsible person linkage are completed. And the inspection efficiency, safety and intelligent level of the equipment in the limited space are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of confined space equipment inspection technology, and in particular to a passive tag-assisted unmanned aerial vehicle (UAV) inspection method and system for key confined space equipment. Background Technology

[0002] Critical equipment such as power distribution cabinets, communication cabinets, fire pumps, and water supply equipment, which are concentrated in confined spaces such as narrow passages, corridors, power distribution rooms, computer rooms, and rooftop equipment platforms within buildings, are prone to overlooking safety hazards. Failures or malfunctions could lead to widespread power outages, water supply disruptions, communication outages, fire protection system failures, and even serious consequences such as casualties. Manual inspection and safety monitoring of critical equipment in confined spaces is prone to accidents and suffers from issues such as missed inspections, incorrect inspections, and incomplete records. Furthermore, while inspection schemes relying on fixed monitoring equipment can collect some data in real time, they are costly to deploy, have limited coverage, and are difficult to flexibly adapt to temporary task requirements. Unmanned Aerial Vehicles (UAVs), with their high mobility, flexible viewing angles, and ability to carry sensors, can quickly enter complex and narrow confined spaces to conduct routine inspections and monitoring of critical equipment, enabling the collection and recording of image and video data.

[0003] However, in confined spaces, Global Navigation Satellite System (GNSS) signals are often severely blocked or subject to severe multipath effects, making it difficult to directly deploy traditional satellite-based UAV inspection technologies. While existing research has attempted to achieve indoor positioning using solutions such as lidar, Simultaneous Localization and Mapping (SLAM), and Visual-Inertial Navigation Systems (VINS), these methods are highly sensitive to environmental features and are prone to positioning drift and cumulative errors when textures are missing or external correction information is lacking. Furthermore, existing methods for detecting critical equipment largely rely on image algorithms to identify equipment nameplates or outlines, which significantly degrades in performance under conditions of strong reflection, complex lighting, or smoke, rain, and fog, resulting in low accuracy in equipment identification. Additionally, there are issues such as a lack of unified data binding and standardized triggering mechanisms for unique equipment identification, spatial location, and specific inspection actions, as well as difficulties in real-time detection of equipment movement and rapid tracing of responsible personnel.

[0004] Radio Frequency Identification (RFID) technology achieves contactless information collection through electromagnetic coupling between a reader and a tag. Currently, most RFID systems use passive tags affixed to the surface of devices, combined with handheld or access control readers for asset inventory and inventory management. This is to balance spatial coverage with reliable identification capabilities. Summary of the Invention

[0005] To address the challenges of drone navigation and positioning in confined spaces, inaccurate inspection results, difficulty in accurately associating inspection data with specific equipment and responsible parties, and the difficulty in timely detection of moved critical equipment, this invention pre-deploys passive RFID tags within the confined space to construct a physical location reference network for equipment and establish a database. RFID codes are then bound to navigation commands, equipment locations, and responsible personnel. During the inspection, the drone reads the tags in real time, triggering navigation commands and inspection tasks. After the inspection, the system compares the detection results with the database information, identifies equipment not in the designated location, and notifies the corresponding responsible person. This achieves precise inspection and location control of critical equipment in confined spaces, thus proposing a passive tag-assisted drone inspection method and system for critical equipment in confined spaces.

[0006] To achieve the above objectives, the present invention provides the following solution: A passive tag-assisted UAV inspection method for critical equipment in confined spaces includes: Construct an RFID tag database, which includes two libraries: a navigation tag database and a device tag database. The confined space of the target is surveyed, a three-dimensional environment model is constructed, and the label installation location is determined; Based on the tag installation location, install RFID tags in the target confined space; Given the limited space of the target after the tag is installed, the UAV uses an RFID identification decision algorithm based on the number of reads and RSSI dual thresholds to perform inspections and analyzes the inspection results.

[0007] Optionally, the navigation tag database includes: RFID codes, action commands, and coordinates of the area where the tag is located; The equipment tag database includes: RFID code, equipment name or number, coordinates or area number of the area where the equipment should be located, name of the corresponding responsible person, contact information of the responsible person, and inspection items that the equipment should perform.

[0008] Optionally, constructing the three-dimensional environment model includes: Using a drone equipped with a LiDAR or RGB-D camera, the confined space of the target is surveyed, and a 3D SLAM algorithm is run to construct a 3D environment model of the confined space, obtaining the geometry of the passage, the turning points, the end of the passage, and the distribution area of ​​equipment.

[0009] Optionally, the label installation location includes: Navigation label data installation locations: at passage bends, passage ends, branch intersections, and danger zone boundaries; Equipment tag data installation location: a conspicuous part of the inspected equipment or a location directly in front of it that is easy to read.

[0010] Optionally, the drone uses an RFID identification decision algorithm based on a dual threshold of read count and RSSI for inspection, including: The drone flies sequentially according to a preset waypoint sequence, periodically calls the radio frequency identification module to scan, and uses an RFID identification decision algorithm based on the number of reads and RSSI dual thresholds to process the scanning results; When the algorithm determines that a tag has been successfully identified and that the tag has a mapping in the navigation tag database or device tag database, a corresponding identification event is generated; Analyze the generated recognition events to complete navigation actions or equipment inspection tasks; After completing navigation actions or equipment inspection tasks, the onboard computing module switches the UAV back to route tracking mode based on the recorded interruption information, restoring the waypoint sequence before the interruption and continuing to execute subsequent waypoint flights.

[0011] Optionally, the RFID identification decision algorithm based on the dual thresholds of read count and RSSI includes: The RFID reader on the drone scans the surrounding tags at fixed intervals and outputs a list of tags read in the current scanning period. Each entry corresponds to a tag, which consists of the tag's unique RFID code, namely the Electronic Product Code Identifier (EPC ID), the Received Signal Strength Indication (RSSI) received by the radio frequency identification module, and the timestamp of this scan. The onboard computing module operates within a set time window. The system caches continuous scan results and categorizes the data by EPC ID. For each EPC ID, count the number of times it is read within the current time window. And calculate the average value of its RSSI. ; Preset read count threshold and RSSI threshold When a certain EPC ID meets the following conditions within the current time window: ,and When the label is successfully identified within the specified time window, it is determined that the label has been successfully identified. Set suppression time for each EPC ID Even if the decision condition is met again during the suppression time, no new recognition event will be output repeatedly. For tags that are successfully identified, the algorithm generates an identification event data packet, including the EPC ID, identification time, and... And the current pose of the drone.

[0012] Optionally, the generated recognition events can be analyzed to complete navigation-related actions or equipment inspection tasks, including: When a navigation tag recognition event is received, the onboard computing module temporarily interrupts the current route tracking according to the action instructions in the navigation tag database, records the waypoint index and flight status before the interruption, and controls the UAV to execute preset actions. When a device tag identification event is received, the onboard computing module controls the drone to hover in front of the device or at a predetermined observation position based on the inspection item list in the device tag database, and calls the vision module to complete image acquisition.

[0013] Optionally, the analysis of the inspection results includes: The actual identification location of each identified device tag is compared with the target area coordinates or area number of the corresponding device in the device tag database. Devices that are recorded in the device tag database but not identified this time, as well as devices whose identification location deviates from the location recorded in the database by more than a preset tolerance threshold, are all identified as location abnormal devices. Based on the name and contact information of the person in charge recorded in the device tag database, alarm records or notification information are generated.

[0014] A passive tag-assisted unmanned aerial vehicle (UAV) inspection system for critical equipment in confined spaces, the system comprising: The tag deployment module is used to pre-deploy passive RFID tags in confined spaces, build a physical location reference network for devices, and establish a database. The airborne computing module is responsible for running RFID identification and decision-making algorithms, UAV inspection and control algorithms, issuing shooting commands, and saving images of key equipment and identified RFID information. The flight control module is responsible for the basic flight control, attitude stabilization, and flight path tracking of the UAV. The radio frequency identification module, including an ultra-high frequency RFID reader and a directional antenna array, is used to scan passive RFID tags deployed in the scene and on the equipment. The communication module is responsible for transmitting all collected tag and device information back to the ground terminal. The vision module is used to take pictures when a shooting command is received.

[0015] The beneficial effects of this invention are as follows: This invention uses passive RFID tags as spatial anchors and mission information sources to achieve low-cost, high-reliability autonomous navigation and equipment inspection in confined space environments with unstable GNSS signals, variable lighting conditions, and complex structures. It dynamically adjusts inspection paths, accurately triggers inspection operations, and completes equipment location verification and personnel linkage, significantly improving the efficiency, safety, and intelligence level of equipment inspection in confined spaces. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This invention provides a framework for a passive tag-assisted unmanned aerial vehicle (UAV) inspection system for critical equipment in confined spaces, as described in an embodiment of the present invention. Figure 2 This is a schematic diagram of a passive tag-assisted UAV inspection method for key equipment in confined spaces according to an embodiment of the present invention. Figure 3 This refers to the navigation tag database and device tag database in this embodiment of the invention. Figure 4 This is the RFID tag identification and decision process according to an embodiment of the present invention; Figure 5 This is an embodiment of the UAV inspection process based on RFID identification event-driven operation. Figure 6 This invention relates to a confined space inspection drone platform. Figure 7 This is a simulated confined space passage and key equipment layout scenario according to an embodiment of the present invention; Figure 8 The drone in this embodiment of the invention identifies a device-type RFID tag; Figure 9 This is the inspection result interface of an embodiment of the present invention; Figure 10 This is the output of the onboard computing module for unmanned aerial vehicle (UAV) identification and navigation RFID tags in this embodiment of the invention. Figure 11This is a diagram showing the change in the forward direction of the drone after it identifies the navigation RFID tag, according to an embodiment of the present invention. Detailed Implementation

[0018] 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.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 2 As shown, this embodiment proposes a passive tag-assisted UAV inspection method for critical equipment in confined spaces, including: Step 1. Construct an RFID tag database, which includes two libraries: a navigation tag database and a device tag database. Step 2. Conduct a survey of the confined space of the target and construct a three-dimensional environment model; Step 3. Determine the label installation location; Step 4. Based on the tag installation location, install RFID tags in the target confined space; Step 5. Based on the confined space of the target after the tag is installed, the UAV uses an RFID identification decision algorithm based on the number of reads and RSSI dual thresholds for inspection; Step 6. Analyze the inspection results.

[0021] Specifically, in this embodiment, step 1: database establishment and tag information entry includes: Encode and input information for all RFID tags, and establish a navigation tag database and a device tag database, such as... Figure 3 As shown. Each navigation tag record includes at least an RFID code (EPC ID), action commands (such as turn 90° left, turn 90° right, ascend 0.5m, descend 0.5m, hover for n seconds, etc.), and the coordinates of the tag's location (such as 3D coordinates or the channel segment number), for subsequent verification. Each device tag record includes at least an RFID code (EPC ID), device name or number, coordinates of the area where the device should be located or area number, the name of the corresponding responsible person, the contact information of the responsible person, and the inspection items that the device should perform (such as taking photos). The above database is stored on the ground server, and the mapping table (RFID code - action command, RFID code - device name and inspection items) is sent to the UAV's onboard computing module.

[0022] Specifically, in this embodiment, step 2: environment modeling, includes: Using drones equipped with LiDAR or RGB-D cameras, the confined space is surveyed, and a 3D SLAM algorithm is run to construct a 3D environment model of the confined space, obtaining the geometry of the passage, turning points, passage ends, and equipment distribution areas. Based on the 3D model, the expected installation locations and heights of various navigation and equipment RFID tags are determined, and the correspondence between the 3D coordinates of these expected locations and the tag codes is recorded in the database of the ground server, providing a spatial reference for subsequent tag deployment and inspection route planning.

[0023] Specifically, in this embodiment, step 3: fixing the RFID tag includes: Based on the 3D environment model obtained in step 2 and the predetermined tag installation locations, affix or fix navigation RFID tags and equipment RFID tags to their respective locations in the confined space: navigation tags are installed on walls or structural surfaces at bends, ends, branch points, and boundaries of hazardous areas; equipment tags are installed in easily readable locations in front of the inspected equipment. During installation, anti-metal or ordinary tags are selected based on the equipment material. After installation, each tag is read using an RFID reader mounted on a drone or a handheld reader to verify that its EPCID matches the database record. The tag position is then fine-tuned based on the actual reading distance, RSSI value, and reading angle. Finally, the actual installation coordinates of each tag are updated in the tag database.

[0024] Specifically, in this embodiment, step 4: RFID tag identification and decision includes: like Figure 4 As shown, considering factors such as multipath reflection, metal obstruction, and tag density in confined spaces, a simple "single-read success" approach can easily lead to false identification. Therefore, this embodiment employs an RFID identification decision algorithm based on a dual threshold of read count and RSSI. The RFID reader scans surrounding tags at fixed intervals (e.g., 20-50ms) and outputs a list of tags read within the current scan period. Each tag contains an EPC ID, RSSI, and a timestamp. The onboard computing module operates within a set time window... The continuous scan results are cached within a short period (e.g., 0.5–1.0 seconds), and the data is categorized by EPC ID. For each EPC ID, the number of times it is read within the current time window is counted. And calculate the average value of its RSSI. Preset read count threshold and RSSI threshold ,For example For 3 to 5 times, It is approximately -60dBm. When a certain EPC ID meets the following conditions within the current time window: ,and When the tag is successfully identified within that time window, it is determined. To prevent the same tag from being repeatedly identified as "successfully identified" within a short period of time, thus triggering the same action repeatedly, the algorithm sets a suppression time for each EPCID. Even if the decision condition is met again within the suppression time, no new recognition event will be output repeatedly. For tags determined to be successfully recognized, the algorithm generates a recognition event data packet, including EPC ID, recognition time, etc. The system records the current drone pose and reports the event to the flight control module and onboard computing module for triggering subsequent action commands or equipment inspection tasks.

[0025] Specifically, in this embodiment, step 5: UAV inspection of confined space, includes: The ground server plans multi-point inspection routes for the UAV based on the environmental model and equipment distribution, and sends the waypoint sequence and basic flight parameters to the UAV's onboard computing module. After takeoff, the UAV enters autonomous inspection mode, using a multi-point route and RFID event interruption collaborative control algorithm combined with the RFID identification algorithm described in step 4 to perform confined space inspection.

[0026] like Figure 5 As shown, the UAV flies sequentially according to a preset waypoint sequence. The flight control module is responsible for attitude stabilization and waypoint tracking. The onboard computing module generates the desired speed and attitude commands in real time based on the spatial relationship between the current position and the next waypoint, and sends them to the flight control module via the communication bus to achieve autonomous flight along the centerline of the channel. During the UAV's flight, the onboard computing module periodically calls the RFID module to scan and processes the scan results using the algorithm described in step 4. When the algorithm determines that a tag has been successfully identified and that the tag has a mapping in the navigation tag database or the device tag database, a corresponding identification event is generated. When a navigation tag identification event is received, the onboard computing module temporarily interrupts the current route tracking according to the action commands in the navigation tag database, records the waypoint index and flight status before the interruption, and controls the UAV to perform actions such as turning, ascending, descending, decelerating, and hovering. When a device tag identification event is received, the onboard computing module controls the UAV to hover in front of the device or at a predetermined observation position according to the inspection item list in the device tag database, and calls the vision module to complete image acquisition. After completing navigation actions or equipment inspection tasks, the onboard computing module switches the UAV back to route tracking mode based on the recorded interruption information, restoring the waypoint sequence before the interruption and continuing to execute subsequent waypoint flights.

[0027] Specifically, in this embodiment, step 6: inspection result analysis, includes: After the inspection, the onboard computing module transmits all RFID codes, identification times, and the drone's pose (which can be approximated as the device's current location or area) of all devices identified during the inspection to the data processing terminal. The actual identification location of each identified device tag is compared with the target area coordinates or area number of the corresponding device in the device tag database. Devices recorded in the database but not identified this time, or devices whose identification location deviates from the database location by a preset tolerance threshold, are identified as having abnormal locations. Based on the responsible person's name and contact information recorded in the device tag database, alarm records or notification messages are generated. The device location verification results and inspection image data are centrally archived for easy browsing and tracing by maintenance personnel.

[0028] This embodiment also provides a passive tag-assisted UAV inspection system for critical equipment in confined spaces, functionally divided into navigation and equipment inspection functions. The system architecture is as follows: Figure 1 As shown. The navigation function is used to enable flight navigation and path guidance for the UAV in confined spaces, while the equipment inspection function is used to identify, acquire images, verify the location, and trace responsibility for key equipment within the confined space.

[0029] The navigation and equipment inspection functions share the same UAV hardware platform. The onboard computing module is responsible for running RFID identification decisions, UAV inspection control algorithms, issuing shooting commands, and saving images of key equipment and identified RFID information. The flight control module is responsible for basic flight control, attitude stabilization, and route tracking. The RFID module includes an Ultra High Frequency (UHF) RFID reader and directional antenna array, specifically designed for scanning passive RFID tags deployed on the scene and equipment. The communication module is responsible for transmitting all collected tag and equipment information back to the ground-based data processing module. The vision module takes pictures when it receives a shooting command.

[0030] Based on the hardware platform, the navigation function is implemented by a waypoint planning module, a positioning module, and a navigation tag database in conjunction with navigation RFID tags. The waypoint planning module generates inspection paths within confined spaces; the positioning module uses VINS positioning to achieve real-time positioning of the UAV, outputting real-time pose and status information; the navigation tag database stores information such as the RFID code, action commands, and area coordinates of the navigation tags. The equipment inspection function is implemented by a data processing module and an equipment tag database in conjunction with equipment RFID tags. The equipment tag database records information such as the equipment RFID code, equipment name, location coordinates, and contact person and contact information; the data processing module compares the actual location of key equipment with the target location in the equipment tag database, identifies equipment not placed in the designated location, and automatically notifies the corresponding responsible person.

[0031] This embodiment proposes a passive RFID tag-assisted UAV inspection method and system for critical equipment in confined spaces. Ordinary or anti-metal tags are affixed to bends in passageways, at the ends of passageways, and on the surface of equipment. Each tag contains a unique EPCID, which is written into both a navigation tag database and an equipment tag database. The RFID code is mapped to information such as "UAV action instructions," "equipment name, equipment location coordinates, and contact information of the responsible person." The UAV, equipped with an Ultra High Frequency (UHF) RFID module and a positioning module, flies along a predetermined waypoint path in the confined space. A decision algorithm based on the number of reads and the Received Signal Strength Indicator (RSSI) is used to stably read the tags, enabling unified triggering of navigation and inspection events. After the inspection, the system automatically compares the detected equipment tags and their actual locations with the target area coordinates recorded in the equipment tag database. It determines equipment that is not placed in the designated location or has been offset, and sends alarms or notifications to the corresponding contact person and contact information in the database. This method uses passive RFID tags as spatial anchors and mission information sources to achieve low-cost, high-reliability autonomous navigation and equipment inspection in confined space environments with unstable GNSS signals, variable lighting conditions, and complex structures. It dynamically adjusts the inspection path, accurately triggers inspection operations, and completes equipment location verification and personnel linkage, significantly improving the efficiency, safety, and intelligence level of equipment inspection in confined spaces.

[0032] Experimental verification in this embodiment: This embodiment uses an unmanned aerial vehicle (UAV) platform and a simulated confined space passage environment to conduct a real-world verification of a passive tag-assisted UAV inspection method for key equipment in confined spaces.

[0033] like Figure 6The image shown is a photograph of the actual UAV hardware platform used in this embodiment. The Pixhawk 6C flight control module is used for attitude stabilization and basic flight control; the NVIDIA Jetson Orin NX onboard computing module is used to run VINS positioning, RFID identification, and task scheduling algorithms; the PRE RFID module is used to read passive RFID tags deployed in the scene; the D435i depth camera module is used to assist in UAV positioning and inspection image acquisition; and the communication module is used for data and command interaction with the ground server.

[0034] like Figure 7 The image shown is a schematic photograph illustrating the simulated confined space passage and key equipment deployment scenario constructed for the actual verification in this embodiment. This scenario includes: ① Simulated passage environment: A narrow, long passage with turns is constructed using partitions, supports, etc., to simulate a confined passage or corridor between buildings. A navigation RFID tag is affixed at the end of the passage to trigger the drone's turning action. ② Key equipment and equipment RFID tags: Key equipment is simulated on one side of the passage using a box, and equipment RFID tags are affixed to prominent positions on the equipment surface. Each equipment tag is registered in the equipment tag database and associated with the equipment number, equipment name, and inspection action. ③ Navigation RFID tags: Navigation RFID tags are affixed at the corners of the passage. These tags are mapped to specific turning commands in the navigation tag database to verify the drone's automatic turning capability after reaching the end of the passage. Through the above deployment, two typical application scenarios are formed: "equipment identification and inspection triggering" and "passage end identification and heading change."

[0035] like Figure 8 The image shown depicts a drone performing an inspection after identifying an RFID tag on equipment in a test scenario. The drone flies into the simulated channel along a preset route, and the RFID module continuously scans surrounding tags at fixed intervals. When the drone reaches a certain distance in front of the equipment, the RFID identification algorithm determines that a particular equipment tag has been successfully identified based on the number of reads and RSSI strength within a time window, and outputs the tag's EPC ID as an identification event to the onboard computing module. The onboard computing module then queries the equipment tag database for the corresponding equipment information and inspection items for that EPC ID, and sends control commands to the flight control module to control the drone to hover in front of the equipment and automatically trigger the vision module to take a picture of the equipment. Figure 8 The data shows that the drone is positioned directly in front of the equipment, with its nose facing the equipment, which meets the observation attitude requirements designed in the database. After recognizing the equipment's RFID tag, the drone is able to correctly interrupt waypoint tracking, perform the predetermined equipment inspection actions, and complete the image acquisition of key equipment.

[0036] like Figure 9The image shown is a screenshot of the inspection results interface. After the drone identifies the equipment's RFID tag and takes a picture, the system automatically reports and archives the inspection results. The inspection results interface displays the equipment's RFID number, inspection time, and the image of the equipment taken by the drone.

[0037] Figure 10 This is an example of the output interface of the onboard computing module after the UAV identifies the navigation RFID tag at the end of the channel. The terminal sequentially displays information such as the tag's EPC code, RSSI, tag detected, and executed turning action, indicating that the system has successfully identified the navigation tag and generated the corresponding turning action command.

[0038] Figure 11 These photos show a drone performing a turning maneuver based on action commands from a navigation tag database in a test scenario. After recognizing the navigation RFID tag at the end of the channel, the drone automatically adjusts its attitude and continues flying along the new channel direction. The change in heading before and after recognition is clearly visible, verifying the RFID-based automatic turning navigation function of this invention.

[0039] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A passive tag-assisted UAV inspection method for critical equipment in confined spaces, characterized in that, include: Construct an RFID tag database, which includes two libraries: a navigation tag database and a device tag database. The confined space of the target is surveyed, a three-dimensional environment model is constructed, and the label installation location is determined; Based on the tag installation location, install RFID tags in the target confined space; Given the limited space of the target after the tag is installed, the UAV uses an RFID identification decision algorithm based on the number of reads and RSSI dual thresholds to perform inspections and analyzes the inspection results.

2. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 1, characterized in that, The navigation tag database includes: RFID codes, action commands, and coordinates of the tag's location. The equipment tag database includes: RFID code, equipment name or number, coordinates or area number of the area where the equipment should be located, name of the corresponding responsible person, contact information of the responsible person, and inspection items that the equipment should perform.

3. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 1, characterized in that, Constructing the three-dimensional environment model includes: Using a drone equipped with a LiDAR or RGB-D camera, the confined space of the target is surveyed, and a 3D SLAM algorithm is run to construct a 3D environment model of the confined space, obtaining the geometry of the passage, the turning points, the end of the passage, and the distribution area of ​​equipment.

4. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 1, characterized in that, The label installation locations include: Navigation label data installation locations: at passage bends, passage ends, branch intersections, and danger zone boundaries; Equipment tag data installation location: a conspicuous part of the inspected equipment or a location directly in front of it that is easy to read.

5. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 1, characterized in that, The drone uses an RFID identification decision algorithm based on both the number of reads and RSSI thresholds for inspection, including: The drone flies sequentially according to a preset waypoint sequence, periodically calls the radio frequency identification module to scan, and uses an RFID identification decision algorithm based on the number of reads and RSSI dual thresholds to process the scanning results; When the algorithm determines that a tag has been successfully identified and that the tag has a mapping in the navigation tag database or device tag database, a corresponding identification event is generated; Analyze the generated recognition events to complete navigation actions or equipment inspection tasks; After completing navigation actions or equipment inspection tasks, the onboard computing module switches the UAV back to route tracking mode based on the recorded interruption information, restoring the waypoint sequence before the interruption and continuing to execute subsequent waypoint flights.

6. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 5, characterized in that, The RFID identification decision algorithm based on the number of reads and RSSI dual thresholds includes: The RFID reader on the drone scans the surrounding tags at fixed intervals and outputs a list of tags read in the current scanning period. Each entry corresponds to a tag, which consists of the tag's unique RFID code (electronic product code), the signal strength received by the radio frequency identification module, and the timestamp of this scan. The onboard computing module operates within a set time window. The system caches continuous scan results and categorizes the data by EPC ID. For each EPC ID, count the number of times it is read within the current time window. And calculate the average value of its RSSI. ; Preset read count threshold and RSSI threshold When a certain EPC ID meets the following conditions within the current time window: ,and When the label is successfully identified within the specified time window, it is determined that the label has been successfully identified. Set suppression time for each EPC ID Even if the decision condition is met again during the suppression time, no new recognition event will be output repeatedly. For tags that are successfully identified, the algorithm generates an identification event data packet, including the EPC ID, identification time, and... And the current pose of the drone.

7. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 5, characterized in that, Analyzing the generated recognition events to complete navigation actions or equipment inspection tasks includes: When a navigation tag recognition event is received, the onboard computing module temporarily interrupts the current route tracking according to the action instructions in the navigation tag database, records the waypoint index and flight status before the interruption, and controls the UAV to execute preset actions. When a device tag identification event is received, the onboard computing module controls the drone to hover in front of the device or at a predetermined observation position based on the inspection item list in the device tag database, and calls the vision module to complete image acquisition.

8. The passive tag-assisted UAV inspection method for critical equipment in confined spaces according to claim 1, characterized in that, The analysis of the inspection results includes: The actual identification location of each identified device tag is compared with the target area coordinates or area number of the corresponding device in the device tag database. Devices that are recorded in the device tag database but not identified this time, as well as devices whose identification location deviates from the location recorded in the database by more than a preset tolerance threshold, are all identified as location abnormal devices. Based on the name and contact information of the person in charge recorded in the device tag database, alarm records or notification information are generated.

9. A passive tag-assisted unmanned aerial vehicle (UAV) inspection system for critical equipment in confined spaces, characterized in that, The system for implementing the method as described in any one of claims 1-8 includes: The tag deployment module is used to pre-deploy passive RFID tags in confined spaces, build a physical location reference network for devices, and establish a database. The airborne computing module is responsible for running RFID identification and decision-making algorithms, UAV inspection and control algorithms, issuing shooting commands, and saving images of key equipment and identified RFID information. The flight control module is responsible for the basic flight control, attitude stabilization, and flight path tracking of the UAV. The radio frequency identification module, including an ultra-high frequency RFID reader and a directional antenna array, is used to scan passive RFID tags deployed in the scene and on the equipment. The communication module is responsible for transmitting all collected tag and device information back to the ground terminal. The vision module is used to take pictures when a shooting command is received.

Images