Electronic fence dynamic protection system based on edge collaborative computing

Abstract

The invention discloses an electronic fence dynamic protection system based on edge collaborative computing, and relates to the field of electric power facility safety protection, the electronic fence dynamic protection system comprises an edge layer, a collaborative layer and a platform layer, the edge layer is deployed around a fence and is used for executing dynamic task allocation according to a network state, analyzing collected surrounding environment information of the fence and sending the analyzed information to the collaborative layer; a corresponding protection strategy is executed according to an analysis result; the collaboration layer is used for constructing network connection among the edge nodes and realizing collaboration among the edge nodes by setting an energy consumption optimizer and a dynamic protection strategy library; and the platform layer is a cloud platform, performs data interaction with the collaboration layer, and is used for providing support for optimization and decision-making of the system and performing remote management and monitoring on the protection system. According to the invention, the safety of the outdoor electric power facility in a severe environment is effectively guaranteed, and an efficient, intelligent and reliable solution is provided for the safety protection of the electric power facility.

Description

Technical Field

[0001] The present invention relates to the field of safety protection of power facilities, and in particular to an electronic fence dynamic protection system based on edge collaborative computing. Background Art

[0002] In the field of power facility security protection, existing technologies have gradually revealed many limitations when dealing with complex scenarios, making it difficult to meet the industry's growing safety and intelligence requirements. The specific limitations are as follows:

[0003] (1) Centralized electronic fence system: Traditional centralized electronic fence systems (such as CN113453142A) adopt an architecture model in which all data collected by sensors is transmitted to a cloud computing center for processing and analysis. Its working principle is that various sensors deployed around the fence, such as infrared sensors and vibration sensors, monitor environmental changes in real time and collect data, which is then uploaded to the cloud computing center via a wireless network. The cloud computing center uses its powerful computing resources to analyze and process the data, determine whether an intrusion or other abnormal situation has occurred, and then feed the processing results back to the fence alarm device.

[0004] However, this architecture has significant drawbacks. For example, during maintenance work at a large hydropower plant, the high-voltage equipment area within the plant is extremely dangerous, and the alarm response time for inadvertent human entry is critical. During a routine maintenance operation, a worker inadvertently approached the high-voltage area. Sensors detected the entry signal and transmitted it to the cloud computing center. However, due to the long data transmission distance and the numerous data requests the cloud computing center must handle, the entire process from sensor data acquisition to processing and alarm signal feedback at the cloud computing center took over two seconds. However, according to hydropower plant maintenance safety regulations, to ensure personnel safety, the alarm response time for inadvertent human entry into the high-voltage area must be within 500ms. This significant time difference prevents traditional centralized electronic fence systems from issuing timely alarms in the face of sudden dangerous situations, seriously threatening personnel safety and equipment operation, and failing to meet the stringent real-time security requirements for power facility protection.

[0005] (2) Wireless networking solution: The currently commonly used wireless networking solution (CN114566016B) mainly uses infrared gratings to construct a protective zone. The working principle of the infrared grating is to install an infrared transmitter and a receiver at each end of the protective zone. The transmitter emits an infrared beam, and the receiver receives the beam to form an invisible protective barrier. When an object blocks the beam, the receiver cannot receive the beam, and the system determines that an intrusion has occurred.

[0006] However, this solution has numerous drawbacks in practical applications. In environments with strong direct sunlight, such as at an open-air substation in a desert area, the intense summer sun can interfere with infrared sensors, causing signal disturbances and misjudgments. According to the substation's operation and maintenance records, during periods of peak sunlight, the infrared gratings reported 5-8 false alarms per hour, severely disrupting staff work and reducing the alarm system's reliability. In foggy and rainy weather, media like mist and dust scatter infrared beams, making it impossible for infrared sensors to accurately determine whether the beam is blocked, similarly leading to a large number of false alarms. In mountainous substations in southern China during the rainy season, this solution experienced a false alarm rate exceeding 45% in foggy and rainy weather.

[0007] Furthermore, this solution's network scale is significantly limited, supporting only 50 nodes at most. For large-scale power facilities, such as ultra-large substations with vast footprints and densely distributed equipment, 50 nodes are far from sufficient for comprehensive coverage, leaving numerous blind spots and making it difficult to meet the needs of large-scale protection. Furthermore, the system utilizes centralized host control. If a host fails, such as a lightning strike at a coastal substation, the entire network will be paralyzed, rendering the protection system ineffective for up to three hours. During these three hours, the substation's protection capabilities are completely lost, and the consequences of a misguided or malicious intrusion during this period would be disastrous.

[0008] Furthermore, this solution lacks a dynamic energy management mechanism, leaving devices like cameras constantly in operation. This results in a single node consuming over 480Wh of power per day. In scenarios where a stable external power source is unavailable, such as outdoor power line protection and substations in remote areas, battery life is severely insufficient. Frequent battery replacement not only increases maintenance costs but also impacts system performance.

[0009] (3) Edge computing network resource sharing architecture

[0010] The closest existing technology, CN202010158429A, proposes an edge computing network resource sharing architecture, attempting to shift computing tasks from cloud computing centers to edge nodes to improve processing efficiency. The basic principle is to deploy edge nodes close to the data collection end. These nodes have a certain level of computing power and can perform preliminary processing and analysis on the data collected by sensors, reducing the amount of data transmitted and the burden on the cloud computing center.

[0011] However, the technology still has unresolved key problems. Its calculation load distribution does not consider network quality factors, and in mountainous substation areas with unstable network signals, high packet loss rates frequently occur. When the edge node needs to migrate the calculation task to other nodes, due to network packet loss, incomplete data transmission, the task migration failure rate exceeds 30%. For example, in mountainous areas under bad weather, the network signal fluctuates greatly, and the task migration request of multiple edge nodes cannot be successfully completed, which seriously affects the overall performance and response speed of the system.

[0012] The technology does not have a hierarchical alarm strategy, and all events trigger sound and light alarms. In actual operation, a large number of false alarms and unnecessary alarm triggers not only cause energy waste, with a sound and light alarm energy consumption ratio of up to 35%, but also make the staff numb to the alarm signal, reducing their sensitivity and vigilance to real dangerous events.

[0013] In addition, the technology does not design a power failure collaborative fault tolerance mechanism. When a node is powered off due to power failure, equipment failure, etc., a security vulnerability will occur in the corresponding protection area, and measures cannot be taken in time to ensure the safety of power facilities. For example, in a sudden power failure accident, part of the nodes stop working, and the system cannot quickly switch or repair, so that the protection area responsible for these nodes loses its protection function and has serious safety hazards. SUMMARY

[0014] To solve the above problems, the purpose of the present application is to provide an electronic fence dynamic protection system based on edge collaborative calculation, which aims to achieve efficient security protection function through collaborative control.

[0015] In order to achieve the above technical purpose, the present application provides an electronic fence dynamic protection system based on edge collaborative calculation, which comprises:

[0016] The edge layer is deployed around the fence and is used to perform dynamic task allocation according to the network state, analyze the collected fence surrounding environment information, and execute the corresponding protection strategy according to the analysis result;

[0017] The collaborative layer is used to build network connection between edge nodes, and through setting energy consumption optimizer and dynamic protection strategy library, the collaborative work between edge nodes is realized;

[0018] The platform layer is a cloud platform, which interacts with the collaborative layer, and is used to provide support for the optimization and decision-making of the system, and remotely manages and monitors the protection system.

[0019] Preferably, the edge layer is also used for distributed task scheduling processing according to task requirements and network conditions, and when the network packet loss rate λ is greater than 0.3, it automatically switches to the local calculation mode to ensure the smooth execution of the task.

[0020] Preferably, the edge layer is further configured to allocate tasks according to the following formula when the network quality is good:

[0021]

[0022] Allocate tasks to different nodes reasonably, where λ is the network packet loss factor, Q is the total number of tasks, and CPU i Indicates the CPU processing capacity of the i-th edge node; CPU k Represents the quantified value of the CPU processing power of the kth edge computing node.

[0023] Preferably, the edge layer also has a fault-tolerant mechanism, which determines the host status by monitoring the node heartbeat signal. When the heartbeat of the master node is lost for more than 5 seconds, the node with the highest CPU idle rate is selected from the slave nodes as the temporary host.

[0024] Preferably, the collaboration layer is also used to build network connections between edge nodes using LoRa wireless communication technology.

[0025] Preferably, the coordination layer is also used to dynamically adjust the working mode of each device according to the system operation status and energy consumption, based on the multi-level power management strategy and dynamic voltage regulation technology.

[0026] Preferably, the collaborative layer is further configured to switch the system to sleep mode when there is no active target and the battery is sufficiently charged; after detecting a target, the device power is gradually increased by the formula Adjust the CPU voltage, where k is the chip process constant, f cpu Dynamically adjusted according to load.

[0027] Preferably, the collaborative layer is also used to call multimodal sensing linkage rules according to different threat levels to achieve collaborative work of sensors, and adaptively switch algorithms based on environmental conditions to improve the environmental adaptability and protection accuracy of the system. In rainy environments, the millimeter wave Doppler algorithm is enabled to accurately detect targets by utilizing the characteristics of millimeter waves penetrating rain and fog; in low-light environments, the low-light vision model YOLOv5s_lowlight is used to enhance image processing and feature extraction to ensure clear target identification even in insufficient light.

[0028] The present invention discloses the following technical effects:

[0029] The present invention effectively ensures the safety of outdoor power facilities in harsh environments and provides an efficient, intelligent and reliable solution for the safety protection of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 is a schematic diagram of the system architecture of the present invention;

[0032] Figure 2 It is a schematic diagram of the design of the system logic function cabinet of the book search of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0034] like Figure 1-Figure 2 As shown, the electronic fence dynamic protection system provided by the present invention adopts a layered architecture design, consisting of an edge layer, a collaboration layer and a cloud platform. Each layer has a clear division of labor and works closely together to achieve efficient security protection functions.

[0035] (1) Edge layer: The edge layer is the frontier of system perception and execution, and includes perception terminals, dynamic computing nodes, and policy executors.

[0036] Perception terminals, deployed around the perimeter of the fence, integrate multiple sensors, such as radar, infrared sensors, and cameras, to collect real-time information about the surrounding environment, including human movement, approaching objects, and environmental parameters such as light intensity, temperature, and humidity. These sensors act as the system's "eyes" and "ears," providing comprehensive and accurate raw data.

[0037] Dynamic computing node: Receives data collected by the sensing terminal. Its built-in distributed task scheduling model and powerful computing capabilities can perform intelligent processing based on task requirements and network conditions. When the network quality is good, according to the task allocation formula (λ is the network packet loss factor, Q is the total number of tasks, CPU iIndicates the CPU processing capacity of the i-th edge node (such as the number of CPU cores, main frequency conversion value or computing power score, which needs to be standardized in advance), reflecting the basic computing resources of the node; CPU k It represents the quantitative value of the CPU processing power of the kth edge computing node, which is used to measure the computing resource "potential" that the node can provide. It is the core indicator for judging the computing power of the node when allocating tasks. It can be understood as follows: among the n edge nodes participating in task allocation, each node has a value representing its own CPU capacity. k This is the "capacity value" of the kth node, such as CPU1 of node 1, CPU2 of node 2, etc., which together constitute The total computing power of all nodes is used to reasonably distribute tasks to different nodes; when the network packet loss rate λ is greater than 0.3, it automatically switches to local computing mode to ensure smooth execution of tasks and effectively avoid task failures caused by network instability.

[0038] Policy Executor: Executes the corresponding protection strategy based on the processing results of the dynamic computing node. For example, when a person is identified as intruder, the sound and light alarm module will be triggered to sound an alarm; when abnormal fence movement is detected, an early warning message will be sent to the relevant personnel. It is the direct executor of the system protection strategy.

[0039] (2) Collaboration layer: The collaboration layer is responsible for achieving collaborative work among edge nodes and consists of a LoRa self-organizing network, an energy consumption optimizer, and a dynamic protection strategy library.

[0040] LoRa ad hoc networks: Utilizing LoRa wireless communication technology, they establish stable network connections between edge nodes. LoRa's long transmission distance, low power consumption, and strong anti-interference capabilities enable large-scale networking of up to 200 nodes. In complex power infrastructure environments, even with widely distributed nodes and challenging terrain, reliable data transmission between nodes is guaranteed, avoiding the limitations of traditional solutions, such as limited network scale and network-wide disruptions caused by host failures.

[0041] Energy Optimizer: monitors the system's operating status and energy consumption in real time, and dynamically adjusts the operating mode of each device based on multi-level power management strategies and dynamic voltage scaling (DVS) technology. For example, when there is no active target and the battery is fully charged, the system switches to sleep mode; when a target is detected, the device power is gradually increased. (k is the chip process constant, f cpu Dynamically adjust the CPU voltage according to load, reducing system power consumption by 52%, effectively solving the problem of unsustainable energy consumption.

[0042] Dynamic Protection Strategy Library: This library stores multiple protection strategies, selects and adjusts them based on actual conditions, and feeds optimized strategies back to the sensing terminal. For example, it invokes multimodal sensor linkage rules based on different threat levels to enable sensor collaboration. It also adaptively switches algorithms based on environmental conditions, improving the system's environmental adaptability and protection accuracy.

[0043] (3) Cloud Platform: The cloud platform plays a core role in management and monitoring in the system. It can send global protection strategies and configuration parameters to the collaborative layer, such as setting threat level standards for different areas and adjusting alarm thresholds. It also receives system operation data and alarm information uploaded by the collaborative layer, performs data analysis and storage, provides support for system optimization and decision-making, and realizes remote management and monitoring of the entire protection system.

[0044] The present invention has an edge collaborative computing architecture: Distributed task scheduling model: The distributed task scheduling model innovatively proposed by the present invention fully considers network quality factors. In practical applications, taking mountain substations as an example, network signals are easily affected by terrain and become unstable. When the network packet loss rate is high, the model can quickly switch to local computing mode and complete the computing task at the local node. For example, in a task of identifying an intruder target, it was originally planned to assign the task to other nodes for calculation, but because the network packet loss rate reached 0.4, the system automatically switched and used edge computing resources at the local node to quickly complete target identification, ensuring the real-time and reliability of the system.

[0045] Fault-tolerance mechanism: This mechanism, implemented in Python code, monitors node heartbeat signals to determine host status. If the master node's heartbeat is lost for more than 5 seconds, the system selects the slave node with the highest CPU idle rate as a temporary host. In a field test at a power facility, simulating a master node power outage due to a fault, the slave nodes detected the heartbeat loss at 5.2 seconds, subsequently elected a temporary host within 200ms, and completed traffic rerouting. System operation was not significantly impacted throughout the entire process, effectively avoiding system paralysis caused by host failure.

[0046] The present invention is equipped with a dynamic protection strategy engine: multimodal sensor linkage rules: the multimodal sensor linkage rules formulated by the system realize the intelligent collaborative work of sensors according to the threat level. In low-risk scenarios, such as daily unmanned areas, the radar scans at a low frequency of 5Hz, the AI ​​module is dormant, and only the LED light flashes at a low frequency to provide a prompt, reducing energy consumption; when a medium-risk situation is detected, such as when a person approaches but no clear intrusion behavior is detected, the radar scanning frequency is increased to 10Hz, the infrared sensor is activated, triggering an 85dB buzzer alarm and recording a log; in high-risk situations, such as when a person forcibly breaks in, the radar scans at a high frequency of 20Hz, the AI ​​recognizes in real time, and the sound and light modules work together to issue a 110dB high-intensity alarm and push information to the cloud to achieve precise protection.

[0047] Adaptive algorithm switching: An adaptive algorithm switching function implemented in C language automatically selects the appropriate algorithm model based on environmental conditions. In rainy environments, the millimeter-wave Doppler algorithm is enabled, leveraging the ability of millimeter-waves to penetrate rain and fog for accurate target detection. In low-light environments, a low-light vision model (such as YOLOv5s_lowlight) is used. Through image enhancement and feature extraction, it ensures clear target recognition even in low light conditions, significantly improving the system's adaptability in complex environments.

[0048] This invention features an intelligent energy management system: a multi-level power management strategy: The system employs a multi-level power management strategy that uses a state machine to switch between different operating modes. For example, in a field power line protection scenario, when there is no intrusion at night and the battery charge drops to 40%, the system automatically shuts off the camera, reduces the radar sampling rate to 2Hz, and extends the interval between audible and visual alarms, switching the system from normal operating mode to energy-saving mode. This reduces power consumption from 185Wh to 62Wh, effectively extending battery life and meeting the requirements for long-term operation without an external power source.

[0049] Dynamic Voltage Scaling (DVS) technology: DVS technology dynamically adjusts the voltage according to the CPU load.

[0050] Achieve precise control. In actual tests, when the system load is low, the CPU frequency is reduced, and the voltage drops accordingly, significantly reducing system power consumption. When the load increases, the voltage and frequency are dynamically adjusted to ensure system performance while achieving energy savings. Tests have shown a 52% reduction in power consumption, improving the system's energy efficiency.

[0051] Through the above technical solution, the present invention effectively solves the defects of the existing technology, achieves major breakthroughs in response delay, network scale, energy consumption management and system reliability, and provides an efficient, intelligent and reliable solution for the safety protection of power facilities.

[0052] Example 1: Maintenance and protection of substation equipment:

[0053] (1) System initialization: Before equipment maintenance work at a large substation, staff members deploy the system. After the first node is powered on, the built-in wireless communication module automatically scans the available frequency bands and selects the 868MHz band to establish a LoRa self-organizing network. This frequency band has good anti-interference capabilities and stable transmission performance in the complex electromagnetic environment of the substation. Based on the preset priority algorithm, the first node determines itself as the host role and begins broadcasting network information.

[0054] After receiving the network signal, the slave node automatically initiates a network access request. After the master node verifies the slave node's identity, it allows it to join the network and assigns it a unique network ID and resource access permissions. After the slave node successfully joins the network, it immediately downloads protection policies tailored to substation maintenance scenarios from the dynamic protection policy library, including intrusion detection rules and alarm threshold settings. Simultaneously, based on the current battery charge (if battery-powered) and environmental parameters, the initial operating mode is set to sleep mode. At this point, except for low-power sensors used to monitor network connectivity and basic environmental parameters, all other devices are in a low-power operating state, reducing the system's initial energy consumption.

[0055] (2) Intrusion response process: When a person or object approaches the maintenance area, the millimeter-wave radar module deployed on the fence first detects a target moving at a speed of 1.5 m / s within 3 m. The radar module transmits the target's distance, speed, direction and other data to node A in real time. After receiving the data, node A calculates its own weight W for processing the task based on the distributed task scheduling model. i , assuming that W is calculated i =0.6, it determines that its own resources are insufficient to efficiently process the task, and therefore sends a computing resource request to the collaboration layer.

[0056] After receiving the request, the collaboration layer traverses the CPU load, memory usage, network connection quality, and other status information of each edge node, selects Node B with a CPU idle rate of 70%, and assigns the task to Node B. After receiving the task, Node B activates the AI ​​model and infrared sensor for joint confirmation. The AI ​​model integrates and analyzes radar data and image data collected by the infrared sensor, identifying target features using a deep learning algorithm. If the AI ​​model determines that the target is an unauthorized person and the recognition confidence level reaches 0.75 (medium risk level standard), the judgment result is transmitted to the policy executor.

[0057] Upon receiving a medium-risk determination, the policy executor immediately triggers the audio and visual module, sounding an 85dB buzzer alarm and recording detailed information about the intrusion in a local log, including the time of occurrence, target location, and the basis for the determination. Node B also uploads this information to the cloud platform via the LoRa network. The cloud platform then maps the intrusion location in real time on an electronic map and pushes an alert to the mobile devices of substation security personnel, prompting them to conduct on-site inspections.

[0058] (3) Energy Optimization Example: When no intrusion occurs at night and the battery level drops to 40%, the energy optimizer comes into play. First, based on the multi-level power management strategy, the energy optimizer sends instructions to each device to shut down the image acquisition and processing functions of all cameras, leaving only the camera's low-power wake-up detection module to detect sudden strong light or large light changes (as a preliminary basis for judging potential intrusion).

[0059] At the same time, the millimeter-wave radar sampling rate was reduced from 10Hz in normal operating mode to 2Hz, reducing the radar's operating frequency and power consumption. For the sound and light alarm module, the alarm interval was extended to 2 seconds to reduce unnecessary power consumption. Through these measures, the overall system power consumption was reduced from 185Wh in normal operating mode to 62Wh, significantly extending battery life and ensuring that the system can continue to operate stably for extended periods of time without an external power source, meeting nighttime safety protection needs.

[0060] Example 2: Fault-tolerance processing in outdoor strong wind environment:

[0061] (1) Failure scenario: A strong typhoon hits the outdoor power facility protection area in a coastal area. The strong wind damages the power supply lines of the host nodes deployed on the fence. The host nodes stop working due to power outage and cannot send heartbeat signals or process data normally, putting the entire protection system at risk of partial failure.

[0062] (2) Collaborative recovery process: The slave node continuously monitors the heartbeat signal of the master node at intervals of 1 second. When the slave node does not receive the heartbeat signal of the master node for 5.2 consecutive seconds, it is determined that the master node has failed. At this time, each slave node begins to compete for the temporary master role according to the preset election algorithm. The election algorithm gives priority to the node with the highest CPU idle rate. Assume that the CPU idle rate of node C reaches 80%, which is higher than that of other slave nodes, and is elected as the temporary master.

[0063] After becoming the temporary master, Node C completes a network configuration update within 200ms, including reallocating network resources and updating routing tables to ensure that other slave nodes can communicate with the temporary master. Subsequently, the temporary master, Node C, reroutes the communication link, redirecting data traffic originally destined for the failed master node to itself, ensuring data transmission continuity.

[0064] At the same time, the temporary host node C synchronizes the system failure and recovery to the cloud platform through the LoRa ad hoc network. After receiving the information, the cloud platform updates the system state display, sends fault alarm and recovery prompt information to the operation and maintenance personnel, so that the operation and maintenance personnel can understand the system operation status in time, and carry out on-site repair when necessary. In the whole fault recovery process, the security protection function of the system does not appear obvious interruption, which effectively guarantees the safety of outdoor power facilities in harsh environment.

[0065] The edge collaborative computing architecture provided by the application has a distributed task scheduling model based on network quality, task dynamic allocation is performed, when the network packet loss rate λ is greater than 0.3, automatic switching to local calculation is realized, the problems of high response delay and task migration failure in the traditional scheme are solved, a fault tolerance mechanism when the node fails is provided, based on heartbeat monitoring, after the main node failure is more than 5 seconds, the node with the highest CPU idle rate is quickly elected as a temporary host, traffic rerouting is realized, and the system reliability is guaranteed.

[0066] The dynamic protection strategy engine provided by the application has a multi-modal sensor linkage rule, different sensor working modes and response actions are set according to the threat level, precise protection is realized, an adaptive algorithm switching mechanism is provided, adaptive algorithm models (such as millimeter wave Doppler algorithm and YOLOv5s_lowlight) are automatically called according to environmental conditions (such as rain, fog and weak light), and the environmental adaptability is improved. The strategy engine is an important innovation and protection content of the application.

[0067] The intelligent energy consumption management system provided by the application has a multi-stage power management strategy, state machine is used to realize mode switching such as sleep, alert and full power, and the dynamic voltage regulation (DVS) technology is combined to dynamically adjust the voltage, so that the system power consumption is reduced by 52%, and the energy consumption problem is effectively solved.

[0068] The communication and system integration provided by the application has a LoRa ad hoc network to realize large-scale stable networking of 200 nodes, and has anti-interference ability; the layered architecture design of the cloud platform, the edge layer and the collaborative layer, and the data interaction and collaborative working mechanism between layers guarantee the efficient operation of the system, and the LoRa-ZigBee dual-mode redundant communication protocol enhances the network reliability.

[0069] The proposed electronic fence dynamic protection system based on edge collaborative computing is applied to the field of power facility security protection. The system constructs an edge layer and collaborative layer architecture, and implements distributed task scheduling and fault tolerance based on network quality through the edge collaborative computing architecture. The dynamic protection strategy engine links multimodal sensors based on threat levels and adaptively switches algorithms. The intelligent energy consumption management system uses multi-level power management and dynamic voltage regulation technology to reduce power consumption. Compared with existing technologies, this solution reduces response delay by 75%, increases network scale by 300%, reduces average daily power consumption by 61.5%, and shortens host fault recovery time by 95%. It effectively solves the problems of slow response, high energy consumption, and low reliability of traditional electronic fences, and provides efficient and reliable technical support for the security protection of power facilities.

[0070] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0071] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0072] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. An electronic fence dynamic protection system based on edge collaborative computing, characterized in that: include: The edge layer is deployed around the fence and is used to perform dynamic task allocation based on network status, analyze the collected environmental information around the fence, and implement corresponding protection strategies based on the analysis results; The collaboration layer is used to build network connections between edge nodes and achieve collaborative work between edge nodes by setting up energy consumption optimizers and dynamic protection strategy libraries; The platform layer is a cloud platform that interacts with the collaboration layer for data, and is used to provide support for system optimization and decision-making, and to remotely manage and monitor the protection system.

2. The electronic fence dynamic protection system based on edge collaborative computing according to claim 1 is characterized by: The edge layer is also used to perform distributed task scheduling according to task requirements and network conditions. When the network packet loss rate λ is greater than 0.3, it automatically switches to local computing mode to ensure smooth execution of tasks.

3. The electronic fence dynamic protection system based on edge collaborative computing according to claim 2 is characterized by: The edge layer is also used to allocate tasks according to the following formula when the network quality is good: Allocate tasks to different nodes reasonably, where λ is the network packet loss factor, Q is the total number of tasks, and CPU i Indicates the CPU processing capacity of the i-th edge node; CPU k Represents the quantified value of the CPU processing power of the kth edge computing node.

4. The electronic fence dynamic protection system based on edge collaborative computing according to claim 3 is characterized by: The edge layer also has a fault-tolerant mechanism that determines the host status by monitoring the node heartbeat signal. When the master node heartbeat is lost for more than 5 seconds, the node with the highest CPU idle rate is selected from the slave nodes as the temporary host.

5. The electronic fence dynamic protection system based on edge collaborative computing according to claim 4 is characterized by: The collaboration layer is also used to build network connections between edge nodes using LoRa wireless communication technology.

6. The electronic fence dynamic protection system based on edge collaborative computing according to claim 5, characterized in that: The collaboration layer is also used to dynamically adjust the working mode of each device according to the system operation status and energy consumption, based on multi-level power management strategy and dynamic voltage regulation technology.

7. The electronic fence dynamic protection system based on edge collaborative computing according to claim 6, characterized in that: The collaborative layer is also used to switch the system to sleep mode when there is no active target and the battery is sufficient; after detecting the target, the device power is gradually increased, through the formula Adjust the CPU voltage, where k is the chip process constant, f cpu Dynamically adjusted according to load.

8. The electronic fence dynamic protection system based on edge collaborative computing according to claim 7 is characterized by: The collaborative layer is also used to call multimodal sensing linkage rules according to different threat levels to achieve collaborative work of sensors. At the same time, it adaptively switches algorithms based on environmental conditions to improve the system's environmental adaptability and protection accuracy. In rainy environments, the millimeter wave Doppler algorithm is enabled to accurately detect targets by utilizing the characteristics of millimeter waves penetrating rain and fog. In low-light environments, the low-light vision model YOLOv5s_lowlight is used to enhance image processing and extract features to ensure clear target identification even in insufficient light.

Images