An electronic fence for radiographic inspection of a nuclear power plant

Abstract

The application discloses a kind of electronic fences for ray flaw detection operation of nuclear power plant, it is related to industrial safety protection field, including the field monitoring terminal, background server system and remote user terminal of intercommunication connection, field monitoring terminal is deployed in the preset isolation boundary of operation site, background server system is as the dispatching hub of entire electronic fence, and remote user terminal is as man-machine interactive interface.The application promotes passive physical isolation to active intelligent electronic fence;Through laser radar and network camera, solve the false alarm and missed alarm problem of single sensor in complex industrial environment;Through background server system, operation, event, image and remote user terminal are associated storage, break data island;Through the bidirectional interaction of remote user terminal and background server system, realize from event trigger, alarm push, artificial disposal to state archiving End-to-end closed loop traceability, so as to meet the management requirements of nuclear power plant to safety protection digitization and refinement.

Description

Technical Field

[0001] This invention relates to the field of industrial safety protection technology, and more specifically, to an electronic fence for radiographic testing operations in nuclear power plants. Background Technology

[0002] During the operation and overhaul of nuclear power plants, radiographic testing is a crucial step in ensuring equipment integrity, but it also carries the risk of high-risk ionizing radiation. Currently, isolation and protection of such high-risk work areas mainly rely on physical warning lines or single security monitoring devices. However, physical isolation lacks proactive sensing and blocking capabilities; single visual monitoring is susceptible to interference and false alarms in industrial settings with complex lighting and intersecting pipelines, and lacks depth information leading to missed alarms; single radar detection, lacking semantic features, struggles to distinguish between human and non-human moving objects. More critically, existing monitoring methods are often independent of the work scheduling system, lacking data correlation between alarm events and specific radiographic testing tasks and on-site equipment status. This results in the inability to achieve end-to-end closed-loop tracing and accurate response in the event of intrusion, failing to meet the digital and refined management requirements of nuclear power plants for safety protection. Summary of the Invention

[0003] In view of this, the present invention discloses an electronic fence for radiographic testing operations in nuclear power plants, in order to solve the problems of large blind spots in single-sensor detection, poor environmental adaptability, and the inability to achieve closed-loop traceability due to the separation of monitoring data and operational business in the existing technology.

[0004] An electronic fence for radiographic testing operations in a nuclear power plant includes: a field monitoring terminal, a back-end server system, and a remote user terminal that are interconnected.

[0005] The back-end server system is used to receive job creation instructions through the remote user terminal, associate the created radiographic testing job with the on-site monitoring terminal deployed at the preset isolation boundary of the job site, and issue an alarm activation instruction to the on-site monitoring terminal when the radiographic testing job is started.

[0006] The on-site monitoring terminal is used to enter the alert state after receiving the alert activation command, and continuously monitor the isolation boundary through integrated lidar and network camera to obtain distance and direction data detected by lidar and monitoring video collected by network camera.

[0007] The on-site monitoring terminal is also used to fuse and analyze the distance and direction data with the monitoring video to identify situations where personnel approach or intrude into the isolation boundary, and generate area intrusion event and on-site image data when an intrusion is determined to have occurred, while triggering its integrated sound and light alarm to provide on-site sound and light warnings.

[0008] The backend server system is also used to acquire the area intrusion event and the on-site image data, push alarm notifications to the remote user terminal, and bind and store the area intrusion event, the on-site image data, the associated X-ray flaw detection operation, and the on-site monitoring terminal for traceability.

[0009] The remote user terminal is used to display the alarm notification and the on-site image data, and to send a processing status update instruction to the back-end server system based on user input.

[0010] Optionally, the backend server system is specifically used for:

[0011] Receive job information containing the job location sent by the remote user terminal, generate the radiographic testing job and assign a job ID;

[0012] Based on the work location, specify the field monitoring terminal deployed at the isolation boundary corresponding to the work location, and obtain the device ID of the field monitoring terminal;

[0013] Bind the job ID to the device ID and set the status of the radiographic testing job to the ready state.

[0014] Optionally, the alert activation command is generated by the background server system when it updates the status of the radiographic testing operation from the "preparing" status to the "working" status after receiving the operation start confirmation;

[0015] When performing the fusion analysis, the on-site monitoring terminal is specifically used for:

[0016] The target points detected by the lidar are converted into spatial coordinate vectors in a three-dimensional Cartesian coordinate system, and the spatial coordinate vectors are projected onto the two-dimensional image pixel coordinate system of the surveillance video based on the extrinsic parameter matrix and the camera intrinsic parameter matrix; the human target detection confidence in the surveillance video is obtained based on the projection position.

[0017] Calculate the target's approach speed and shortest radial distance relative to the isolation boundary based on the distance and direction data;

[0018] A comprehensive threat index is calculated based on the detection confidence level, the approach speed, and the shortest radial distance. An intrusion event is determined and generated only when the comprehensive threat index is greater than a preset judgment threshold and the shortest radial distance is less than or equal to zero.

[0019] Optionally, the on-site monitoring terminal is further specifically used to automatically control the network camera to capture on-site images that trigger the intrusion event when the intrusion event is generated, as the on-site image data, or to store the on-site images that trigger the intrusion event to a specified directory for the background server system to download as the on-site image data.

[0020] The backend server system is also specifically used to push alarm notifications containing event type, location and time of occurrence to the remote user terminal responsible for the radiographic testing operation in real time.

[0021] Optionally, the backend server system has a built-in database cluster, which includes job tables, device tables, and event tables. When performing binding storage, the backend server system is specifically used for:

[0022] Assign an event ID to the intrusion event in the area, and persistently store the event ID, event type, occurrence time, occurrence location, and image path of the scene image data in the event table;

[0023] Establish association fields with the job ID and the device ID in the event table to build an association data model to support combined querying and tracing of historical events by time, location, job, or event type.

[0024] Optionally, the remote user terminal is specifically used to receive the user's confirmation operation on the alarm notification, generate the processing status update instruction, and send it to the backend server system;

[0025] The backend server system is also specifically used to update the processing status of the area intrusion event in the event table to "processing completed" according to the processing status update instruction; and to continuously monitor the online status of the on-site monitoring terminal through a heartbeat mechanism, and when the on-site monitoring terminal is determined to be offline, generate an offline event record and trigger an offline alarm.

[0026] Optionally, the remote user terminal includes an office terminal or mobile terminal that accesses the back-end server system through an encrypted communication channel;

[0027] The backend server system is built on a distributed microservice architecture and specifically includes device management services, job scheduling services, and event processing services.

[0028] The device management service is used to monitor the device status of the field monitoring terminal.

[0029] The job scheduling service is used to manage the radiographic testing operations and their status.

[0030] The event handling service is used to handle intrusion events in the area and control the real-time push of alarm notifications.

[0031] Optionally, the on-site monitoring terminal is a floor-standing, portable all-in-one device, which includes, from bottom to top, a base assembly, a connecting rod, and a host control assembly.

[0032] The base assembly is equipped with casters with brakes at the bottom, and a battery compartment is provided on the base assembly for independently powering the field monitoring terminal.

[0033] The connecting rod is vertically mounted on the base assembly, and the host control assembly is mounted on the top of the connecting rod.

[0034] Optionally, the front end of the host control component is provided with a display screen and the network camera, and the network camera is located below the display screen;

[0035] The main control component is provided with protruding warning lights and sound units on both sides, and the warning lights and sound units together constitute the sound and light alarm.

[0036] The lidar is mounted at the top center of the host control component, and the lidar protrudes from the upper surface of the host control component, such that the horizontal scanning surface of the lidar is higher than the upper surface of the host control component.

[0037] Optionally, the front end face of the host control component is further provided with a host power switch and a lidar power switch;

[0038] The external casing of the field monitoring terminal adopts a casing structure that meets the preset protection and explosion-proof level, and the field monitoring terminal integrates a communication module for accessing the back-end server system through the factory's local area network.

[0039] As can be seen from the above technical solution, this invention discloses an electronic fence for X-ray flaw detection operations in nuclear power plants, including a field monitoring terminal, a back-end server system, and a remote user terminal that are interconnected. The field monitoring terminal is deployed at a preset isolation boundary at the work site, the back-end server system serves as the scheduling hub for the entire electronic fence, and the remote user terminal serves as the human-machine interface. This invention upgrades passive physical isolation to an active intelligent electronic fence through the active perception and audible and visual warnings of the field monitoring terminal; it solves the problem of false alarms and missed alarms of single sensors in complex industrial environments by using the fusion analysis of lidar and network cameras and the complementarity of depth information and semantic features; it breaks down data silos by associating and storing operations, events, images, and remote user terminals through the back-end server system; and it achieves end-to-end closed-loop traceability from event triggering, alarm push, manual handling to status archiving through bidirectional interaction between the remote user terminal and the back-end server system, thereby meeting the requirements of nuclear power plants for digital and refined management of safety protection. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the published drawings without creative effort.

[0041] Figure 1 This is a topological diagram of the overall architecture of an electronic fence for radiographic testing in a nuclear power plant, as disclosed in an embodiment of the present invention.

[0042] Figure 2 This is an overall physical structure diagram of a field monitoring terminal disclosed in an embodiment of the present invention;

[0043] Figure 3 This is a partial schematic diagram of a host control component and sensor layout disclosed in an embodiment of the present invention;

[0044] Figure 4 This is a flowchart of a multi-source sensor fusion and regional intrusion identification algorithm disclosed in an embodiment of the present invention;

[0045] Figure 5 This is a schematic diagram illustrating the spatial coordinate system mapping principle between a lidar and a network camera, as disclosed in an embodiment of the present invention.

[0046] Figure 6 This is a timing diagram of the life cycle and multi-terminal collaborative control of a flaw detection operation disclosed in an embodiment of the present invention;

[0047] Figure 7This is a schematic diagram of a microservice architecture and associated data model of a backend server system disclosed in an embodiment of the present invention. Detailed Implementation

[0048] 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.

[0049] To address the problems of large blind spots, poor environmental adaptability, and lack of closed-loop traceability caused by the separation of monitoring data from operational tasks in existing technologies, this application proposes an electronic fence for radiographic testing operations in nuclear power plants, which achieves strict spatial alignment of heterogeneous sensor data and digital safety management and control throughout the entire lifecycle of the operation.

[0050] See Figure 1 The present invention discloses an overall architecture topology diagram of an electronic fence for radiographic testing operations in a nuclear power plant. The electronic fence for radiographic testing operations in a nuclear power plant includes: a field monitoring terminal 10, a back-end server system 20, and a remote user terminal 30 that are interconnected.

[0051] Among them, the on-site monitoring terminal 10 is deployed as an edge sensing node at the preset isolation boundary of the work site, the back-end server system 20 is deployed as the core scheduling hub in the data center or cloud, and the remote user terminal 30 serves as a human-machine interaction window for management personnel. It should be understood that, although Figure 1 The specific topological connection relationship between the three is shown in the figure, but in other embodiments, any data link that can realize the following functional interaction is within the protection scope of the present invention. For example, the field monitoring terminal 10 can access the back-end server system 20 through various means such as wired industrial Ethernet and 5G private network to adapt to the network infrastructure conditions of different areas in the nuclear power plant.

[0052] The backend server system 20 is used to receive job creation instructions through the remote user terminal 30, associate the created radiographic testing job with the field monitoring terminal 10 deployed at the preset isolation boundary, and issue an alarm activation instruction to the field monitoring terminal 10 when the radiographic testing job is started.

[0053] In this embodiment, the backend server system 20 is built on a distributed microservice architecture and specifically includes device management services, job scheduling services, and event processing services. The backend server system 20 is not merely a passive data storage center, but also the logical control core of the entire security protection system. Upon receiving a job creation instruction containing information such as job location and time, the system automatically selects the field monitoring terminal 10 that best fits the isolation boundary in its physical location based on a preset spatial matching algorithm, and establishes a unique mapping relationship between the virtual job task and the physical monitoring equipment. This association mechanism ensures that every high-risk X-ray inspection job has one and only corresponding dedicated electronic fence for protection, avoiding the risk of mismatch or omission that may occur due to manual configuration. When the job status switches from the preparation stage to the start stage, the backend server system 20 immediately generates an alert activation instruction and sends it to the associated field monitoring terminal 10, driving the front-end device to seamlessly switch from a low-power standby mode to a high-precision real-time monitoring mode, thereby achieving automated synchronization of the security management process and the equipment operating status.

[0054] The on-site monitoring terminal 10 is used to enter an alert state after receiving the alert activation command, and continuously monitor the isolation boundary through integrated lidar and network camera to obtain distance and direction data detected by the lidar and monitoring video collected by the network camera.

[0055] In this embodiment, the on-site monitoring terminal 10 integrates a heterogeneous sensor array, utilizing the high-precision ranging capability of lidar and the rich semantic acquisition capability of webcam to construct a multi-dimensional environmental perception foundation. LiDAR can acquire precise three-dimensional spatial information of targets within isolated boundary areas regardless of lighting conditions, while the webcam provides texture, color, and morphological features of the targets. The synchronous acquisition of data in both timestamps and spatial coordinate systems provides high-quality raw input for subsequent data fusion. This dual-modal perception design effectively overcomes the limitations of single sensors in complex industrial environments, maintaining stable detection performance even at night or under direct sunlight.

[0056] The on-site monitoring terminal 10 is also used to fuse and analyze the distance and direction data with the monitoring video to identify situations where personnel approach or intrude into the isolation boundary, and to generate area intrusion event and on-site image data when an intrusion is determined to have occurred, while triggering its integrated sound and light alarm to provide on-site sound and light warning.

[0057] It is important to note that the fusion analysis here is a functional processing step. Its core lies in spatiotemporally aligning and cross-validating the geometric spatial information provided by radar with the visual semantic information provided by video, rather than being limited to a specific mathematical formula or algorithm model. In practical implementation, fusion analysis can include, but is not limited to, combinations of various techniques such as point cloud projection based on coordinate transformation, target detection confidence weighting based on deep learning, and trajectory prediction based on kinematic features. Through this fusion mechanism, the system can effectively distinguish between genuine personnel intrusions and interference caused by on-site equipment movement, animal passage, or changes in lighting, significantly reducing the false alarm rate. Once a substantial intrusion is confirmed, the on-site monitoring terminal 10 will trigger a local audible and visual alarm within milliseconds, using high-decibel voice prompts and high-frequency flashing lights to immediately deter and dissuade the intruder. This front-end autonomous response mechanism does not rely on network transmission latency, maximizing personnel safety in high-risk radiation environments. Simultaneously, the on-site monitoring terminal 10 reports the generated area intrusion event and on-site image data to the back-end server system 20, providing a solid chain of evidence for back-end tracing and handling. In addition, the on-site monitoring terminal 10 can also send heartbeat data packets containing its own device ID to the host computer at a fixed frequency.

[0058] The background server system 20 is also used to acquire the area intrusion event and the on-site image data, push alarm notifications to the remote user terminal 30, and bind and store the area intrusion event, the on-site image data, the associated X-ray flaw detection operation, and the on-site monitoring terminal 10 for traceability.

[0059] Specifically, upon receiving alarm information reported from the front end, the backend server system 20 immediately executes two parallel operations: one is to precisely push the alarm notification, including the event type, location, time, and on-site snapshot, to the remote user terminal 30 responsible for the operation via real-time communication protocols such as message queues, ensuring that management personnel can be informed of the on-site situation immediately; the other is to write the alarm data to the database cluster and, using previously established relationships, forcibly bind the intrusion event with the specific operation ID and device ID. This deeply bound storage strategy breaks down the information silos between alarm records and business data in traditional security systems, enabling post-event audits to quickly retrieve the complete context of the violation event through any dimension (such as operation number, device number, or time period), providing tamper-proof digital evidence for security incident investigation, evidence collection, and management optimization.

[0060] The remote user terminal 30 is used to display the alarm notification and the on-site image data, and to send a processing status update instruction to the back-end server system 20 based on user input.

[0061] The remote user terminal 30 can be an office computer installed in the central control room, or an explosion-proof tablet or smartphone carried by the safety officer, as long as it can access the backend server system 20 through an encrypted communication channel. On the display interface, the terminal not only presents alarm text information, but also intuitively displays on-site images captured at the moment the alarm is triggered, assisting managers in quickly determining the nature of the event.

[0062] In summary, this invention discloses an electronic fence for X-ray flaw detection operations in nuclear power plants, comprising a field monitoring terminal 10, a back-end server system 20, and a remote user terminal 30, all interconnected. The field monitoring terminal 10 is deployed at a pre-defined isolation boundary at the work site. The back-end server system 20 serves as the scheduling hub for the entire electronic fence, and the remote user terminal 30 acts as the human-machine interface. This invention elevates passive physical isolation to an active intelligent electronic fence through the proactive sensing and audible / visual warnings of the field monitoring terminal 10. By fusing the analysis of LiDAR and network cameras, and utilizing the complementarity of depth information and semantic features, it solves the problem of false alarms and missed alarms caused by single sensors in complex industrial environments. The back-end server system 20 associates and stores operations, events, and images with the remote user terminal 30, breaking down data silos. Through bidirectional interaction between the remote user terminal 30 and the back-end server system 20, it achieves end-to-end closed-loop traceability from event triggering, alarm push notifications, manual handling, to status archiving, thereby meeting the requirements of nuclear power plants for digital and refined management of safety protection.

[0063] In one embodiment, the on-site monitoring terminal 10 is a floor-standing, portable all-in-one device, which includes, from bottom to top, a base assembly 11, a connecting rod 12, and a host control assembly 13.

[0064] Specifically, such as Figure 2 As shown, the on-site monitoring terminal 10 is a floor-mounted, mobile, all-in-one device. This integrated vertical structure design gives the device a high degree of integration and independence, enabling it to be deployed as a complete sensing and interaction node in the complex industrial site of a nuclear power plant.

[0065] The base assembly 11 serves as the supporting foundation for the entire device, and its bottom is equipped with casters with brakes. This design gives the on-site monitoring terminal 10 the ability to move flexibly within the factory area. Workers can easily push it to any temporary work area or open space lacking fixed monitoring infrastructure. Upon reaching the preset isolation boundary, locking the brakes ensures the device remains firmly in place, effectively preventing displacement due to accidental collisions or ground slopes, and ensuring the physical stability of the electronic fence boundary definition. It should be understood that while this embodiment preferably uses omnidirectional casters with brakes, in other embodiments, pneumatic tires, rubber tracks, or magnetic mobile bases can be used depending on the site conditions, as long as the device can be deployed mobilely and locked at a fixed point.

[0066] The base assembly 11 also includes a battery compartment for independently powering the on-site monitoring terminal 10. Specifically, considering that nuclear power plant X-ray inspection areas are often located in temporary locations such as under-construction buildings, basements, or equipment mezzanines, these areas typically lack readily available mains power interfaces or have extremely difficult access to power cables. The built-in high-capacity battery compartment (e.g., using lithium iron phosphate or ternary lithium battery packs) effectively eliminates the equipment's reliance on external cables, enabling truly wireless deployment. This not only significantly shortens the preparation time for on-site deployment but also avoids the risk of tripping or damage caused by power cables crossing passageways. In a preferred embodiment, the battery compartment can adopt a modular quick-release design, supporting rapid on-site battery pack replacement for uninterrupted operation, thus meeting the needs of high-intensity continuous operation during overhauls.

[0067] The connecting rod 12 is vertically mounted on the base assembly 11, and the main control assembly 13 is mounted on top of the connecting rod 12. The connecting rod 12 not only provides physical support but also raises the main control assembly 13 to a suitable height, such as between 1.5 and 2 meters, ensuring a wide horizontal field of view for the LiDAR and network camera, while also facilitating on-site personnel to observe display information and perform switching operations. Structurally, the connecting rod 12 can be a fixed-length metal tube or a telescopic and adjustable sleeve structure to adapt to the varying monitoring height requirements of different operating scenarios. The main control assembly 13, as the core functional compartment of the equipment, integrates an edge computing unit, sensor interface board, and power management module internally, while externally housing various human-machine interface and sensor components.

[0068] To optimize multimodal perception and human-computer interaction performance, the external layout of the host control component 13 is specifically designed. In one embodiment, a display screen and a webcam are mounted on the front surface of the host control component 13, with the webcam located below the display screen. Figure 3As shown, this vertical arrangement conforms to ergonomic visual habits. The display screen is positioned above, allowing personnel to easily view the work status, alarm information, or debugging parameters while standing. The network camera, located below the display screen, better covers the near-ground blind spots in front of the equipment, while also preventing glare interference from direct light from the display screen onto the lens. The main control component 13 has protruding warning lights and sound units on both sides, forming an audible and visual alarm. Placing the audible and visual alarm on the sides of the device, rather than the front or back, achieves 360-degree omnidirectional warning coverage. When an intrusion is detected, regardless of the direction from which personnel approach the equipment, they will immediately receive a high-frequency flash and a high-decibel voice prompt, eliminating potential visual or auditory blind spots associated with one-way alarms. The protruding structural design also helps to prevent the casing from obstructing sound wave propagation and light radiation, improving the penetration of the warning signal.

[0069] A key structural feature of this embodiment lies in the installation method of the lidar. As one implementation method, to ensure an absolutely wide detection field of view, as shown in the attached diagram... Figure 3 As shown in the top spatial layout, a lidar is installed in the center of the top space of the host control component 13, and the lidar protrudes from the upper surface of the host control component 13, making the horizontal scanning plane of the lidar higher than the upper surface of the host control component 13. Under this spatial layout, a local spatial geometric constraint model needs to be established. The horizontal plane containing the upper surface of the host control component is defined as the reference plane, and its equation is: The coordinates of the transmission center of the lidar are defined as follows: Its horizontal scanning surface equation is .

[0070] Specifically, such as Figure 3 As shown, the horizontal plane containing the upper surface of the host control component 13 is defined as the reference plane, and its height coordinate is... The elevation coordinates of the LiDAR's transmission center (i.e., the center of the optical window) are defined as follows: This invention strictly satisfies geometric constraints through the protruding design of the mechanical structure. This physical constraint ensures that the detection beam emitted by the lidar, within its effective detection field of view, has its corresponding horizontal scanning surface completely higher than the upper surface of the host control component. This physically eliminates the obstruction of the detection beam by the device's own casing, acquiring blind-spot-free ambient point cloud data, thus laying the foundation for subsequent acquisition of accurate distance and direction data. Therefore, this seemingly simple dimensional relationship has extremely important technical significance: it physically ensures that the lidar's emitted beam is not obstructed by the top cover of the host control component 13's casing during 360-degree horizontal rotation scanning. If the lidar is installed too low or flush with the casing, the scanning beam will be truncated by the device itself in the near-field region, forming an undetectable blind spot, causing intrusions near the device to be missed. Through this protruding design, the system acquires truly omnidirectional point cloud data without blind spots, providing complete and reliable raw input for high-precision fusion analysis. It should be understood that the protrusion is not limited to directly raising the mounting base. It can also be achieved by setting a groove on the top of the host and embedding the radar while ensuring that the light window is higher than the edge of the groove, or by using a cantilever bracket to suspend the radar. As long as the functional requirement of unobstructed horizontal scanning surface is met, it falls within the protection scope of this invention.

[0071] Furthermore, to adapt to the stringent industrial environment and safety management regulations of nuclear power plants, this invention also incorporates targeted designs for equipment operation and protection. As one implementation method, the front face of the main control component 13 is also equipped with a main power switch and a lidar power switch, enabling independent physical control of the core computing unit and the high-power detection unit. This dual-switch independent control design allows operators to independently cut off the power supply to the high-power lidar during non-operational periods or equipment maintenance, while retaining the main unit's low-power standby or communication functions. This extends battery life and avoids unnecessary laser consumption and potential safety risks during unnecessary periods. Simultaneously, the independent physical switches facilitate on-site fault isolation and troubleshooting, improving operational efficiency.

[0072] As attached Figure 3 The spatial geometric reference of the heterogeneous sensor shown below establishes the laser radar transmission center. With the optical core of the network camera The relative spatial translation vector between :

[0073] ;

[0074] The translation vector The mechanical dimensions of the host control component are factory-fixed. This design, which integrates the LiDAR and webcam onto the same rigid structure, ensures that their relative spatial pose remains absolutely static. This constant spatial constraint provides a stable and unique physical reference for coordinate system calibration and spatial mapping transformation during multi-sensor data fusion analysis, ensuring the alignment accuracy of radar distance data and 2D surveillance video over long-term operation.

[0075] In one embodiment, the front face of the host control component is further provided with a host power switch and a lidar power switch.

[0076] The external casing of the field monitoring terminal 10 adopts a casing structure that meets the preset protection and explosion-proof level, and the field monitoring terminal 10 integrates a communication module for accessing the back-end server system 20 through the factory's local area network.

[0077] Specifically, some work areas in nuclear power plants may contain dust, water vapor, oil, and even flammable and explosive gases. Therefore, the enclosure must meet at least an IP65 dustproof and waterproof rating and satisfy the corresponding Ex explosion-proof certification requirements depending on the area. The enclosure material can be corrosion-resistant engineering plastics, aluminum alloy, or stainless steel, with surface treatment for antistatic and anti-radiation aging to ensure structural integrity and electrical safety even under long-term harsh environments. The on-site monitoring terminal 10 integrates a communication module for accessing the backend server system 20 via the plant's local area network. This communication module supports multiple access methods such as Wi-Fi, 4G / 5G, or industrial Ethernet, and has a built-in high-gain antenna to adapt to the complex electromagnetic shielding environment within the plant. After power-on, the communication module automatically searches for and registers with the plant's private network, eliminating the need for manual IP address configuration or wiring, achieving plug-and-play functionality and rapid network setup. This highly integrated hardware design makes the field monitoring terminal 10 not only a sensing front end, but also an industrial-grade intelligent terminal that is highly adaptable, reliable, and easy to deploy, perfectly meeting the comprehensive requirements of nuclear power plant X-ray flaw detection operations for the mobility, durability, and intelligence of safety protection equipment.

[0078] In one embodiment, the backend server system 20 can be specifically used for:

[0079] Receive job information including job location sent by the remote user terminal 30, generate the radiographic testing job and assign a job ID;

[0080] Based on the work location, specify the field monitoring terminal 10 deployed at the isolation boundary corresponding to the work location, and obtain the device ID of the field monitoring terminal 10;

[0081] Bind the job ID to the device ID and set the status of the radiographic testing job to the ready state.

[0082] This embodiment constitutes the necessary enabling condition for triggering the subsequent high-precision fusion algorithm. Specifically, the system does not run in a high-power fusion monitoring mode throughout all time periods. Instead, it establishes a unique mapping relationship between virtual business tasks and physical sensing devices through precise binding of job IDs and device IDs. Only when a specific X-ray flaw detection job is created and associated with a specific field monitoring terminal 10 does that terminal have the logical basis to enter an alert state. This design effectively avoids invalid calculations and false alarms caused by environmental interference during non-operational periods or in irrelevant areas, while also ensuring on-demand allocation of algorithm resources and improving the overall energy efficiency and business relevance of the system.

[0083] Based on this, the alert activation command is generated by the backend server system 20 when it updates the status of the X-ray flaw detection operation from the "preparing" state to the "working" state after receiving the operation start confirmation. In other words, simply creating the operation and binding it to the equipment is not enough to trigger monitoring; a clear start confirmation signal must be issued by the operator or administrator, and the system state machine must complete the transition from "preparing" to "working" before issuing the alert activation command. This state gating mechanism further eliminates interference from frequent personnel entering and exiting to deploy equipment during the operation preparation phase, ensuring that the electronic fence is only effective during the window period when radiation risks actually exist.

[0084] When performing fusion analysis, the on-site monitoring terminal 10 can accurately identify personnel intrusion behavior. For example... Figure 4 and Figure 5 As shown, this fusion analysis process mathematically couples the three-dimensional spatial perception capability of LiDAR with the two-dimensional semantic recognition capability of webcam.

[0085] The on-site monitoring terminal 10 is specifically used for:

[0086] (1) Convert the target point detected by the lidar into a spatial coordinate vector in a three-dimensional Cartesian coordinate system.

[0087] (2) Project the spatial coordinate vector onto the two-dimensional image pixel coordinate system of the surveillance video based on the extrinsic matrix and the camera intrinsic matrix; obtain the human target detection confidence in the surveillance video based on the projection position.

[0088] In practical applications, for the current frame of the surveillance video provided by the network camera, the spatial coordinate vector is mapped across dimensions based on the extrinsic parameter matrix and the camera intrinsic parameter matrix. The spatial coordinate vector is then projected onto the two-dimensional image pixel coordinate system of the surveillance video, and the projected pixel coordinates are calculated.

[0089] In the current frame of the monitoring video, the on-site monitoring terminal 10 extracts a preset neighborhood image region centered on the projected pixel coordinates, and inputs the rain image of the preset neighborhood image region into the visual semantic recognition model to calculate the detection confidence of the presence of human targets.

[0090] (3) Calculate the approach speed of the target relative to the isolation boundary and the shortest radial distance based on the distance and direction data.

[0091] The on-site monitoring terminal uses multiple consecutive frames of lidar detection data and a time-difference algorithm to calculate the approach speed of the target relative to the preset isolation boundary. And extract in real time the shortest radial distance of the target from the isolation boundary. .

[0092] (4) Calculate the comprehensive threat index based on the detection confidence, the approach speed and the shortest radial distance. If and only if the comprehensive threat index is greater than the preset judgment threshold and the shortest radial distance is less than or equal to zero, determine that a person has broken in and generate the area intrusion event.

[0093] This invention constructs a regional intrusion threat index assessment model, processes the detection confidence, approach speed, and shortest radial distance to calculate the comprehensive threat index, and obtains the comprehensive threat index.

[0094] An intrusion event is determined and a region intrusion event is generated only when the comprehensive threat index is greater than a preset judgment threshold and the shortest radial distance is less than or equal to zero. That is, it is determined that a person in a moving state has substantially crossed the isolation boundary and a region intrusion event is generated. Otherwise, the initial state is returned and multi-threaded monitoring continues.

[0095] For details, please see Figure 4 and Figure 5 As shown, in order to achieve accurate identification and intrusion determination of moving personnel and their spatial location, this invention combines the attached... Figure 4 The flowchart of the multi-source sensor fusion and regional intrusion detection algorithm shown below, along with the attached... Figure 5 The diagram shown illustrates the principle of spatial coordinate system mapping between LiDAR and network camera. This invention elaborates on the underlying computational logic of how the on-site monitoring terminal fuses distance and direction data with monitoring video for analysis.

[0096] As attached Figure 4 As shown in the process starting point, after receiving the alarm activation command issued by the back-end server system 20, the on-site monitoring terminal 10 enters the alarm state, and its internal host control component reads the point cloud stream of the lidar and the video stream of the network camera in parallel through multi-threading.

[0097] First, in conjunction with the appendix Figure 5 The coordinate system mapping process constructs a three-dimensional local spatial polar coordinate system based on the lidar transmission center. The relative distance of the detected target point in this polar coordinate system is obtained through hardware scanning. and horizontal deflection To achieve spatial alignment of heterogeneous sensor data, homogeneous coordinate transformation is used to convert the polar coordinate data into a spatial coordinate vector in a three-dimensional Cartesian coordinate system. The specific mathematical expression of this coordinate transformation process is as follows:

[0098] ;

[0099] in, , , These represent the spatial displacement components of the target point detected by the lidar along three mutually perpendicular coordinate axes in a three-dimensional Cartesian coordinate system. The number 1 indicates a homogeneous coordinate term, used to support subsequent affine transformation operations. Since this embodiment limits the lidar to horizontal plane scanning, this local coordinate system... The value is set to zero.

[0100] Subsequently, as attached Figure 5 As shown in the joint calibration calculation step, a cross-dimensional spatial mapping relationship is established between the three-dimensional local polar coordinate system where the LiDAR is located and the two-dimensional image pixel coordinate system where the network camera is located. Through the joint calibration calculation operation at the factory stage, the extrinsic parameter matrix of the network camera relative to the LiDAR is obtained. (It should be noted that this extrinsic parameter matrix) In a mathematical and physical sense, it integrates the aforementioned spatial translation vector determined by the mechanical structure. and the relative rotation matrix obtained through calibration calculation ), and the camera intrinsic parameter matrix of the webcam itself. Among them, the camera intrinsic parameter matrix Based on the extrinsic parameter matrix With camera intrinsic parameter matrix The two-dimensional projection formed by the two elements will transform the previously obtained spatial coordinate vectors. The optical center of the network camera is directly projected onto the two-dimensional image pixel coordinate system of the surveillance video, and the corresponding projected pixel coordinates are calculated. According to the projected pixel coordinates The depth distance value can be obtained, and the specific projection mapping calculation equation is as follows:

[0101] ;

[0102] in, To detect the depth and distance values ​​of the target in the camera coordinate system, and These represent the horizontal pixel column coordinates and vertical pixel row coordinates mapped onto the two-dimensional monitoring image, respectively.

[0103] Based on the above rigorous physical space mapping, returning to the appendix... Figure 4 The algorithm flowchart shows that the on-site monitoring terminal 10 accurately extracts the projected pixel coordinates from the current video frame of the monitored video. A preset neighborhood image region is defined as the center point, and the image data of this region is input into a built-in visual semantic recognition model (such as a deep learning-based YOLO series object detection model, Faster R-CNN model, or SSD model; in this embodiment, to balance the computing power limitations and real-time performance of edge computing devices, a lightweight deployment of the YOLOv5s object detection network model is preferred). The detection confidence of the presence of a human target in this region is then calculated. Simultaneously, the on-site monitoring terminal utilizes multiple consecutive frames of lidar detection data to calculate the approach speed of the target relative to the preset isolation boundary based on a time-difference algorithm. And extract in real time the shortest radial distance of the target from the isolation boundary. .

[0104] This invention constructs a regional intrusion threat index assessment model, which mathematically integrates the aforementioned multi-dimensional spatial motion features with visual semantic features. (See attached image.) Figure 4 As shown in the threat index calculation section, this model is used to calculate the comprehensive threat index of a target in real time. Its calculation formula is defined as:

[0105] ;

[0106] in, To approximate the kinematic weighting coefficients related to the velocity and boundary distance, For visual recognition confidence weighting coefficients; It is a preset small normal number used to prevent program overflow exceptions where the target is exactly on the boundary line and the denominator is zero when the underlying computing unit performs division operations.

[0107] By continuously calculating this comprehensive threat index The on-site monitoring terminal establishes a set of multi-constraint intrusion judgment criteria. (See attached...) Figure 4 As shown in the decision box, the comprehensive threat index is determined if and only if the calculated comprehensive threat index is... The shortest radial distance is greater than the preset threshold and is calculated as follows. When the system determines that a person is moving and their spatial location substantially crosses the isolation boundary into the danger zone, an intrusion is detected, and an intrusion event is generated. This fusion analysis mechanism utilizes the strong coupling constraint relationship between spatial location, kinematic features, and visual semantic confidence to eliminate false alarms from single sensors caused by birds, fallen leaves, dust obstruction, or equipment vibration in industrial settings, ensuring the absolute reliability of alarm triggering.

[0108] In one embodiment, the remote user terminal includes an office terminal or mobile terminal that accesses the back-end server system via an encrypted communication channel.

[0109] The backend server system is built on a distributed microservice architecture and specifically includes device management services, job scheduling services, and event processing services.

[0110] To support the high concurrency and high reliability business needs within nuclear power plants, the back-end server system abandoned the traditional monolithic application design and instead adopted a microservice decoupled architecture.

[0111] The device management service is responsible for communicating with the underlying hardware to monitor the device status of the field monitoring terminal, including heartbeat maintenance and offline determination.

[0112] The job scheduling service independently undertakes the workflow logic of the business layer and is used to manage the radiographic testing job and its status.

[0113] The event handling service is used to handle intrusion events in the area and control the real-time push of alarm notifications.

[0114] This architecture allows each functional module to be deployed and scaled independently. When a service (such as event processing) experiences a sudden surge in load due to a large number of alarms, it will not affect the normal dispatching of job scheduling services, thus ensuring the overall stability of the system. Simultaneously, the remote user terminal 30 uses an encrypted communication protocol with the backend, ensuring the secure transmission of sensitive operational data and security video within the nuclear power plant's intranet environment. It should be understood that although this embodiment lists three core microservices, in other implementations, log auditing, user access control, etc., can also be separated into independent services based on business scale. As long as the distributed decoupling design principle is followed, they all fall within the scope of this invention.

[0115] like Figure 6 The flaw detection operation lifecycle and multi-terminal collaborative control timing diagram shown in this embodiment are described in detail to realize the digital collaborative management and control of the entire lifecycle of radiographic flaw detection operations, and to describe the specific execution logic of multi-terminal collaboration, state machine transition and event closed-loop control.

[0116] As attached Figure 6Phase 1 involves pre-allocation of resources and physical anchoring. The backend server system first receives business layer job information containing the job location through the remote user terminal. After parsing the job information, the system automatically generates a task instance for this radiographic testing job and assigns it a unique job ID within the system's global scope.

[0117] Subsequently, the backend server system extracts the job location coordinates from the job information. It also iterates through the deployment coordinates of currently available field monitoring terminals recorded in the preset database. To achieve precise matching and anchoring between physical boundaries and virtual tasks, the backend server system calculates the two-dimensional Euclidean distance between the job location and the deployment locations of each device. Its calculation formula is defined as:

[0118] ;

[0119] in, , These are the local planar projection coordinates of the center location of the radiographic testing operation. , These are the projected coordinates corresponding to the actual deployment location of the on-site monitoring terminal. Calculated by the system. The system selects the minimum on-site monitoring terminal that meets the isolation boundary coverage requirements and retrieves its corresponding device ID. After spatial matching is completed, the system forcibly binds the job ID and the device ID at the data table level, and simultaneously sets the initial value of the state machine for the radiographic testing job to the ready state.

[0120] As attached Figure 6 Phase Two, as shown in the state torque and underlying alert activation process, involves the backend server system receiving a job start confirmation command from a remote user terminal via network monitoring after completing the aforementioned preliminary resource allocation and binding. This command triggers the backend core control flow, updating the status of the X-ray inspection operation from the "preparing" state to the "working" state. Along with this state machine attribute reversal, the backend server system synchronously generates the alert activation command and sends it to the bound field monitoring terminal via the plant's dedicated network, driving its underlying sensor array and fusion analysis logic into continuous monitoring mode.

[0121] As attached Figure 6In Phase Three, during the event triggering and manual intervention phase, when the on-site monitoring terminal determines that a person's spatial location has entered the isolation boundary and generates an intrusion event based on the aforementioned fusion analysis logic, its internal microcontroller automatically controls the network camera to perform a hardware-level snapshot. The system captures an on-site image at the exact moment the intrusion event is triggered as the on-site image data. The on-site monitoring terminal encodes this image or persistently stores it in a preset local directory for download by the backend server system via a file transfer protocol as the on-site image data.

[0122] After obtaining the complete event context, the backend server system pushes an alarm notification to the remote user terminal responsible for the X-ray inspection operation in real time. The alarm notification data packet strictly encapsulates the event type, the location determined based on the equipment deployment coordinates, and the occurrence time with a timestamp accurate to milliseconds. The remote user terminal renders and displays the alarm notification and corresponding on-site image data on its screen interface, assisting management personnel in quickly assessing the on-site situation.

[0123] Based on the user's screen interaction input, the remote user terminal generates a processing status update instruction and sends it back to the backend server system. The backend server system parses the network packet and, according to the event identifier carried in the packet, overwrites and updates the processing status field of the area intrusion event in the underlying database to "processing completed," thereby forming a complete state machine closed loop from event response to manual intervention.

[0124] In addition, as attached Figure 6 The parallel heartbeat online assessment mechanism shown in Phase 4, operating at a fixed frequency, ensures the absolute reliability of the online status of edge devices throughout the entire business process. The backend server system continuously monitors the network connection status of the field monitoring terminals using a heartbeat mechanism. Each field monitoring terminal sends heartbeat data packets containing its own device ID to the host computer at a fixed frequency. The backend server system records the system arrival timestamps of two consecutive successful receptions of heartbeat data packets from the same source. and And calculate the communication time interval. :

[0125] ;

[0126] in, The system's receiving time for the latest heartbeat packet. This is the system reception time of the last valid heartbeat packet. When the system determines the calculated... If the communication timeout exceeds the preset threshold, the monitoring terminal is directly determined to be abnormally offline. At this time, the system automatically generates an offline event record and triggers a targeted device offline alarm to the remote user terminal, effectively eliminating the silent monitoring blind spot caused by power depletion or network interruption.

[0127] Once the operation is started and the monitoring mode is entered, if an intrusion occurs, the system will enter the event response phase. As one implementation method, the on-site monitoring terminal is also specifically used to automatically control the network camera to capture images of the scene when the intrusion event is triggered, as the on-site image data, or to store the images of the scene when the intrusion event is triggered to a designated directory for download by the backend server system as the on-site image data.

[0128] The backend server system is also specifically used to push alarm notifications containing event type, location and time of occurrence to the remote user terminal responsible for the radiographic testing operation in real time.

[0129] Specifically, the timing of the image capture is strictly synchronized with the moment the fusion algorithm determines the intrusion, ensuring the timeliness of the evidence chain. Considering potential network fluctuations in industrial settings, the system supports two image acquisition modes: directly uploading the encoded image stream when the network is good; and persisting the image to a designated local directory during network congestion, with the image then asynchronously downloaded by the backend server system via a file transfer protocol once the network recovers. The alarm notification data packet strictly encapsulates the event type (e.g., "Level 1 Intrusion"), the location determined based on the device deployment coordinates, and the occurrence time with a timestamp accurate to milliseconds, allowing managers to quickly assess the severity of the situation without physically being present at the site.

[0130] To enable complete tracing of historical events, the backend server system has a built-in database cluster, which includes job tables, device tables, and event tables. Specifically, during binding and storage, the backend server system assigns event IDs to intrusion events in a given area and persistently stores the event ID, event type, occurrence time, occurrence location, and image path of the on-site image data in the event table. It also establishes association fields with job IDs and device IDs in the event table, constructing a relational data model to support combined queries and tracing of historical events by time, location, job, or event type.

[0131] In one embodiment, the remote user terminal is specifically used to receive the user's confirmation operation on the alarm notification, generate the processing status update instruction, and send it to the backend server system;

[0132] The backend server system is also specifically used to update the processing status of the area intrusion event in the event table to "processing completed" according to the processing status update instruction; and to continuously monitor the online status of the on-site monitoring terminal through a heartbeat mechanism, and when the on-site monitoring terminal is determined to be offline, generate an offline event record and trigger an offline alarm.

[0133] See Figure 7 The diagram illustrates the microservice architecture and associated data model of the backend server system. In this embodiment, to achieve reliable storage of massive amounts of business data and high cohesion and low coupling of system functions, this invention details the specific implementation logic of the backend server system built on a distributed microservice architecture and database cluster. Under this architecture, the backend server system abandons the traditional monolithic application design, decomposing the overall business logic and deploying it as independently running microservice processes.

[0134] Specifically, the backend server system, based on a distributed microservice architecture, includes device management services, job scheduling services, and event handling services. The device management service is specifically responsible for communicating with the underlying hardware, processing and monitoring the device status information of the on-site monitoring terminals, including the aforementioned heartbeat maintenance and offline analysis. The job scheduling service independently handles the workflow logic of the business layer, managing the lifecycle of the X-ray flaw detection operation and its status transitions. The event handling service, as the core response node for abnormal situations, handles the reception and parsing of intrusion events and directly controls the real-time push of alarm notifications to the corresponding remote user terminals.

[0135] To support the data persistence requirements of the aforementioned microservice modules, the backend server system has a built-in database cluster with high availability. The underlying data structure of the database cluster includes a job table, a device table, and an event table. When the event processing service receives a reported intrusion information, the backend server system, during the binding and storage process, first assigns a globally unique event ID to the intrusion event. Subsequently, the system packages and persists the event ID, event type, occurrence time accurate to milliseconds, location coordinates bound to the device, and the absolute image path of the aforementioned on-site image data in the server's file system into the event table.

[0136] Building upon the physical storage of individual events, the backend server system establishes association fields with job IDs and device IDs in the event table. By introducing foreign key constraints from relational or graph databases, the system tightly anchors isolated alarm records to specific flaw detection tasks and hardware entities, thereby constructing a three-dimensional relational data model at the underlying level.

[0137] Based on the relational data model, the backend server system established a multi-dimensional historical event tracing algorithm. The system defines the condition vector for query requests. Each component represents a retrieval constraint in the time domain, spatial location domain, task identifier domain, and event type domain, respectively. Simultaneously, the feature vector of a single historical record in the event table is extracted as follows: The system constructs a combined query matching function. The calculation formula is:

[0138] ;

[0139] in, Feature index representing the query dimension . For each dimension, there is an independent Boolean discriminant function; if the query request does not specify the first dimension... When constraints are imposed on a dimension (i.e., the dimension is allowed to have any value), or features with historical records where such constraints are imposed. The discriminant function value of this dimension when the constraint interval is fully satisfied. Forced assignment of 1; only when the first value is explicitly applied. Dimensional constraints, but features If this condition is not met, determine the value of the discriminant function. Only 0.

[0140] If and only if the system traversal calculation yields At that time, determine the historical record. The system successfully retrieves the requested data. Through this rigorous filtering mechanism based on vectorized logical multiplication, the system allows administrators to perform precise combined queries and tamper-proof tracing of massive historical events by time, location, task, or event type via remote user terminals, achieving a complete data closed loop from triggering and response to post-event auditing of nuclear power plant safety events.

[0141] In summary, the electronic fence for radiographic testing in nuclear power plants provided by this invention has the following beneficial effects:

[0142] This invention effectively overcomes the blind spots in traditional single-vision monitoring when targets move radially, achieving multi-dimensional spatial perception. Conventional area intrusion alarms on the market often rely on planar image analysis, which is prone to missed detections when personnel move perpendicular to the camera. This invention deeply integrates distance and direction data detected by LiDAR with monitoring video from network cameras through a field monitoring terminal. This not only endows the system with the ability to accurately capture moving personnel at any location and distance within the detection range, but also effectively eliminates interference from non-human targets using visual semantics, significantly improving the accuracy of intrusion detection in complex industrial environments.

[0143] This invention constructs a three-dimensional safety response mechanism that transitions from passive warning to proactive defense. Relying on the communication link between the backend server system and field equipment, when a field terminal enters a state of alert and determines that personnel are abnormally approaching, it can instantly trigger an integrated audible and visual alarm to physically deter and dissuade people from entering the area. Simultaneously, the system pushes the generated area intrusion event and on-site image data to remote user terminals in real time. This coordinated intervention of "immediate front-end blocking + synchronous back-end sensing" significantly reduces the risk of accidental exposure in high-risk radiation environments.

[0144] This invention establishes a digital closed-loop traceability system based on multi-dimensional data association. The system's built-in database cluster no longer stores alarm information in isolation, but automatically binds each captured on-site image, its precise time and geographical location, to the currently executing X-ray inspection operation and the corresponding on-site monitoring terminal. This processing logic, which transforms discrete security alarms into structured business data, supports subsequent panoramic retrieval and auditing based on time, location, or operation type, providing objective and detailed structured data evidence for incident review and safety management.

[0145] This invention features high system availability and strong business scalability. The backend server system is built on a distributed microservice architecture, modularizing and decoupling core functions such as equipment management, job scheduling, and event handling, smoothly integrating the previously complex X-ray inspection process into a digital workflow. Combined with the support of an underlying database cluster and a continuous heartbeat disconnection monitoring mechanism, the system ensures smooth status transitions and reliable alarm pushes even when facing complex factory network fluctuations or multi-point concurrent operations.

[0146] This invention provides a flexible hardware deployment solution that is highly adaptable to harsh industrial environments. The on-site monitoring terminal innovatively adopts a floor-standing, mobile, all-in-one design with integrated braked casters at the bottom. It features an independent power supply battery compartment and a local area network communication module, effectively eliminating the dependence of traditional security equipment on on-site cable installation and power supply conditions. Combined with a casing structure that meets explosion-proof and high protection standards, the device can be pushed to the isolation boundary of an open area and automatically connect to the system network and accept dispatch upon power-up, perfectly addressing the challenges of temporary, high-risk operations in the variable spaces and harsh environments of nuclear power plants.

[0147] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0148] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0149] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An electronic fence for X-ray flaw detection operations in nuclear power plants, characterized in that, include: The on-site monitoring terminals, back-end server system, and remote user terminals are interconnected and communicate with each other. The back-end server system is used to receive job creation instructions through the remote user terminal, associate the created radiographic testing job with the on-site monitoring terminal deployed at the preset isolation boundary of the job site, and issue an alarm activation instruction to the on-site monitoring terminal when the radiographic testing job is started. The on-site monitoring terminal is used to enter the alert state after receiving the alert activation command, and continuously monitor the isolation boundary through integrated lidar and network camera to obtain distance and direction data detected by lidar and monitoring video collected by network camera. The on-site monitoring terminal is also used to fuse and analyze the distance and direction data with the monitoring video to identify situations where personnel approach or intrude into the isolation boundary, and generate area intrusion event and on-site image data when an intrusion is determined to have occurred, while triggering its integrated sound and light alarm to provide on-site sound and light warnings. The backend server system is also used to acquire the area intrusion event and the on-site image data, push alarm notifications to the remote user terminal, and bind and store the area intrusion event, the on-site image data, the associated X-ray flaw detection operation, and the on-site monitoring terminal for traceability. The remote user terminal is used to display the alarm notification and the on-site image data, and to send a processing status update instruction to the back-end server system based on user input.

2. The electronic fence for radiographic testing in nuclear power plants according to claim 1, characterized in that, The backend server system is specifically used for: Receive job information containing the job location sent by the remote user terminal, generate the radiographic testing job and assign a job ID; Based on the work location, specify the field monitoring terminal deployed at the isolation boundary corresponding to the work location, and obtain the device ID of the field monitoring terminal; Bind the job ID to the device ID and set the status of the radiographic testing job to the ready state.

3. The electronic fence for radiographic testing in nuclear power plants according to claim 2, characterized in that, The alert activation command is generated by the background server system when it updates the status of the radiographic testing operation from the "preparing" state to the "working" state after receiving the operation start confirmation. When performing the fusion analysis, the on-site monitoring terminal is specifically used for: The target points detected by the lidar are converted into spatial coordinate vectors in a three-dimensional Cartesian coordinate system, and the spatial coordinate vectors are projected onto the two-dimensional image pixel coordinate system of the monitoring video based on the extrinsic matrix and the camera intrinsic matrix. The confidence level of human target detection in the surveillance video is obtained based on the projection position; Calculate the target's approach speed and shortest radial distance relative to the isolation boundary based on the distance and direction data; A comprehensive threat index is calculated based on the detection confidence level, the approach speed, and the shortest radial distance. An intrusion event is determined and generated only when the comprehensive threat index is greater than a preset judgment threshold and the shortest radial distance is less than or equal to zero.

4. The electronic fence for radiographic testing in nuclear power plants according to claim 3, characterized in that, The on-site monitoring terminal is also specifically used to automatically control the network camera to capture on-site images that trigger the intrusion event when the intrusion event is generated in the area, as the on-site image data, or to store the on-site images that trigger the intrusion event in the area to a specified directory for the background server system to download as the on-site image data. The backend server system is also specifically used to push alarm notifications containing event type, location and time of occurrence to the remote user terminal responsible for the radiographic testing operation in real time.

5. The electronic fence for radiographic testing in nuclear power plants according to claim 4, characterized in that, The backend server system has a built-in database cluster, which includes job tables, device tables, and event tables. Specifically, the backend server system is used for binding storage as follows: Assign an event ID to the intrusion event in the area, and persistently store the event ID, event type, occurrence time, occurrence location, and image path of the scene image data in the event table; Establish association fields with the job ID and the device ID in the event table to build an association data model to support combined querying and tracing of historical events by time, location, job, or event type.

6. The electronic fence for radiographic testing in nuclear power plants according to claim 5, characterized in that, The remote user terminal is specifically used to receive the user's confirmation of the alarm notification, generate the processing status update instruction, and send it to the backend server system. The backend server system is also specifically used to update the processing status of the area intrusion event in the event table to "processing completed" according to the processing status update instruction; and to continuously monitor the online status of the on-site monitoring terminal through a heartbeat mechanism, and when the on-site monitoring terminal is determined to be offline, generate an offline event record and trigger an offline alarm.

7. The electronic fence for radiographic testing in nuclear power plants according to claim 1, characterized in that, The remote user terminal includes an office terminal or mobile terminal that accesses the back-end server system through an encrypted communication channel. The backend server system is built on a distributed microservice architecture and specifically includes device management services, job scheduling services, and event processing services. The device management service is used to monitor the device status of the field monitoring terminal. The job scheduling service is used to manage the radiographic testing operations and their status. The event handling service is used to handle intrusion events in the area and control the real-time push of alarm notifications.

8. The electronic fence for radiographic testing in nuclear power plants according to claim 1, characterized in that, The on-site monitoring terminal is a floor-standing, portable all-in-one device, which includes, from bottom to top, a base assembly, a connecting rod, and a main control assembly. The base assembly is equipped with casters with brakes at the bottom, and a battery compartment is provided on the base assembly for independently powering the field monitoring terminal. The connecting rod is vertically mounted on the base assembly, and the host control assembly is mounted on the top of the connecting rod.

9. The electronic fence for radiographic testing in nuclear power plants according to claim 8, characterized in that, The front end of the host control component is provided with a display screen and the network camera, and the network camera is located below the display screen; The main control component is provided with protruding warning lights and sound units on both sides, and the warning lights and sound units together constitute the sound and light alarm. The lidar is mounted at the top center of the host control component, and the lidar protrudes from the upper surface of the host control component, such that the horizontal scanning surface of the lidar is higher than the upper surface of the host control component.

10. The electronic fence for radiographic testing in nuclear power plants according to claim 9, characterized in that, The front face of the host control component is also provided with a host power switch and a lidar power switch. The external casing of the field monitoring terminal adopts a casing structure that meets the preset protection and explosion-proof level, and the field monitoring terminal integrates a communication module for accessing the back-end server system through the factory's local area network.

Images