A teaching environment security guarantee method based on edge computing

Abstract

This invention discloses a teaching environment security assurance method based on edge computing, including step one: three-layer node deployment and localized collection of multi-source data; step two: layered edge analysis and accurate identification of security events; step three: graded early warning and real-time linkage and response at the edge; step four: edge-cloud collaboration and iterative optimization of global strategies; and step five: early warning escalation and closed-loop response. This invention, by constructing a three-layer edge computing architecture consisting of regional edge nodes, campus edge center nodes, and cloud collaborative nodes, achieves closed-loop management of the entire process of "perception-analysis-early warning-response-optimization" for teaching environment security events. It comprehensively covers the security needs of all teaching scenarios, including teaching buildings, laboratories, and dormitories, effectively protecting the personal safety of teachers and students, the safety of campus property, and maintaining normal campus teaching order.

Description

Technical Field

[0001] This invention relates to the field of teaching environment security technology, and in particular to a teaching environment security method based on edge computing. Background Technology

[0002] The teaching environment, as the core setting for teachers and students' learning and activities, encompasses diverse areas such as teaching buildings, laboratories, libraries, student dormitories, and training centers. Its safety is directly related to the personal safety of teachers and students, the safety of campus property, and the normal teaching order. Currently, the safety of the teaching environment mostly adopts a cloud-based centralized computing model. After collecting data such as videos, environment, and personnel through various sensing devices, all of them are uploaded to the cloud server for analysis and processing before issuing early warnings and handling instructions.

[0003] This model faces several unresolved problems in practical applications: First, it suffers from poor real-time performance and high response latency. Safety incidents in the teaching environment (such as laboratory chemical leaks, student falls, dormitory fires caused by illegal electrical use, and stampedes caused by crowds gathering in corridors) are sudden, and the long-distance transmission of high-definition video streams and real-time data from multiple sensors is easily affected by network bandwidth and congestion. The process of cloud analysis and feedback often misses the golden time for response. Second, it carries a high risk of privacy leaks and high bandwidth costs. The teaching environment includes video data from private areas such as dormitories and restrooms. Uploading all of this data to the cloud poses a data leakage risk, and the continuous transmission of massive amounts of raw data also poses a risk of data leakage. The data will consume the core network bandwidth of the campus, significantly increasing the operation and maintenance costs; third, the system has low fault tolerance and is prone to overall paralysis. The cloud server is a single core node. Once a failure or network outage occurs, the entire security system will lose its ability to analyze and handle the situation and will be unable to perform any security monitoring; fourth, the system has poor scenario adaptability and homogenized analysis. The safety needs of different areas of the teaching environment vary significantly (e.g., laboratories focus on the safety of hazardous chemicals and equipment, dormitories focus on the safety of electricity and personnel, and playgrounds focus on the safety of personnel activities). The generalized analysis model in the cloud cannot achieve differentiated and accurate identification based on the characteristics of each area, resulting in low accuracy of early warning.

[0004] Currently, although edge computing technology has been initially applied in the security field, enabling local analysis of some data, existing solutions are mostly applicable to scenarios such as commercial parks and residential communities. However, they lack a systematic edge computing architecture design and methodology to address the diverse scenarios, specific personnel, and standardized emergency response procedures required in teaching environments. This prevents the complementary advantages of rapid local processing at the edge and global overall optimization at the cloud, and also fails to design a collaborative mechanism for tiered early warning and layered response for teaching scenarios. Consequently, it cannot effectively address the pain points of traditional cloud-based models in ensuring the security of teaching environments. Therefore, this invention proposes a teaching environment security method based on edge computing to solve the problems existing in the prior art. Summary of the Invention

[0005] To address the aforementioned issues, this invention aims to propose a teaching environment security assurance method based on edge computing. By constructing a three-layer edge computing architecture consisting of regional edge nodes, campus edge center nodes, and cloud collaborative nodes, it achieves local collection, layered analysis, and real-time response of multi-source security perception data in the teaching environment, solving problems such as response delays, privacy leaks, and high bandwidth consumption inherent in traditional cloud-based models. Differentiated edge analysis models and early warning strategies are designed for the security characteristics of different areas within the teaching environment, enabling accurate identification and tiered handling of security incidents. Simultaneously, a dynamic collaborative mechanism between the edge and cloud sides is established, enabling global data storage, model iteration, and cross-regional collaborative scheduling through the cloud. This enhances the fault tolerance and self-optimization capabilities of the teaching environment security assurance system, ultimately achieving a closed-loop management of the entire process of "perception-analysis-early warning-handling-optimization" for teaching environment security incidents, comprehensively protecting the personal safety of teachers and students and maintaining campus teaching order.

[0006] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a teaching environment security assurance method based on edge computing, implemented using a three-layer edge computing node architecture consisting of regional edge nodes, campus edge center nodes, and cloud collaborative nodes, comprising the following steps: Step 1: Divide the teaching area into protected zones according to the security level, and deploy edge nodes in each zone; Step 2: Perform real-time analysis of the data using a lightweight analysis model, then identify security events and label them with regional warning levels. Through the campus edge center node, perform cross-regional correlation analysis on abnormal feature data, identify composite risks, and label them with campus-level warning levels. Step 3: Implement early warning measures based on the regional and campus-level early warning levels; Step 4: The campus edge center node encrypts and uploads the entire process data of the security incident to the cloud collaborative node for data storage and model training, and iteratively optimizes the lightweight analysis model and the teaching scenario security feature library. Step 5: Secondary assessment of regional early warnings and handling of escalated incidents to form a closed-loop management system.

[0007] Further improvements are made in the following: In step one, the regional edge node integrates multi-source sensing devices and a lightweight analysis module with a built-in lightweight analysis model to collect and preprocess local multi-source sensing data for subsequent analysis. Among them, core data involving privacy are stored only locally, while abnormal feature data is uploaded to the campus edge center node. The data preprocessing includes format conversion, invalid data removal, and redundant data compression.

[0008] A further improvement is that the multi-source sensing device includes one or more of the following: a high-definition camera, an environmental sensor, an equipment status sensor, a personnel density sensor, and an access control sensing terminal.

[0009] The further improvement lies in the following: In step one, the division of the protection zone specifically divides the teaching environment into a core protection zone, a general protection zone, and an outer protection zone. The core protection zone includes laboratories, student dormitories, hazardous chemical warehouses, and training centers. The general protection zone includes classrooms in teaching buildings, libraries, and canteens. The outer protection zone includes campus corridors, playgrounds, and recreational areas.

[0010] Further improvements are made in the following aspects: In step two, the identification of safety events is specifically carried out in conjunction with a pre-set knowledge base of safety characteristics of teaching scenarios. The knowledge base of safety characteristics of teaching scenarios pre-stores characteristic parameters and judgment thresholds corresponding to different teaching areas and different types of safety events, and supports custom settings and updates; The regional warning levels include level one warning, level two warning and level three warning. Level one warning corresponds to emergency events that require immediate action, level two warning corresponds to dangerous events, and level three warning corresponds to general events.

[0011] Further improvements are made in the following aspects: In step three, the early warning response is specifically led by the edge node of the region that triggered the early warning, based on the regional early warning level, to coordinate the security equipment in the region and issue response instructions to the designated management personnel. Based on the campus-level early warning level, the campus edge center node leads the cross-regional equipment linkage and personnel coordination and scheduling.

[0012] Further improvements are made in the following aspects: When handling the early warning, the handling of the first-level early warning includes the activation of the sound and light alarm by the regional edge node, the shutdown of the corresponding valve or power supply, the activation of the sprinkler or smoke exhaust device, and the push of early warning information containing the event location, type and real-time video to the handheld terminal of the regional management personnel. The response to a Level 2 alert includes coordinating with nearby high-definition cameras to locate the target, sending alert information to security personnel and area administrators, and initiating on-site investigation and interception procedures. The handling process for Level 3 early warnings includes sending an early warning notification only to the area administrator and initiating on-site evacuation / equipment inspection procedures.

[0013] Further improvements include: during the campus-level early warning and response, each edge node collects response feedback data in real time and synchronizes it to the campus edge center node to achieve dynamic tracking of the response process. The response feedback data includes the arrival time of security personnel, on-site response measures, and the development status of the event.

[0014] The further improvement lies in the following: In step four, the cloud-based collaborative nodes iteratively optimize the lightweight analysis model based on a large database of historical security events. The model is trained using machine learning algorithms, and the optimized model parameters are then distributed to edge nodes in each region for local updates.

[0015] The further improvement is that: Step five specifically means that if the regional warning is not handled within a preset time or the event escalates, it will be automatically reported to the campus edge center node for secondary evaluation, and the cloud collaborative node will be triggered to intervene to expand the scope of handling until the event is handled and confirmed. It also includes the provision that when the campus edge center node or the regional edge node fails or loses network access, the cloud collaborative node temporarily takes over the analysis and scheduling tasks to ensure that the system is not interrupted.

[0016] The beneficial effects of this invention are as follows: By constructing a three-layer edge computing architecture, this invention realizes the localized collection and hierarchical analysis of safety perception data in the teaching environment, which greatly reduces the identification and response delay of safety incidents, solves the problem of delayed response to sudden safety incidents, and firmly grasps the golden time for handling. By storing core privacy data locally and only uploading data with abnormal characteristics, the risk of cloud leakage of teachers' and students' privacy data is avoided from the source. At the same time, the bandwidth consumption of the campus network is greatly reduced, and the operation and maintenance costs of the campus security system are significantly reduced. To address the varying safety needs across different areas of the teaching environment, a differentiated, lightweight analysis model and tiered response strategy were designed. This enabled accurate identification and targeted handling of safety incidents, significantly improving the accuracy of early warnings and resolving the issue of poor scenario adaptability of traditional generalized analysis models. Through the dynamic edge-cloud collaboration mechanism, the advantages of rapid edge-side processing and global optimization on the cloud side are complemented. The cloud completes iterative optimization of models and strategies based on massive data and distributes them to the edge side, enabling the entire security system to have self-learning and self-optimization capabilities, which greatly improves the stability of the system and avoids the paralysis of the entire system caused by the failure of a single node. This invention realizes a closed-loop management of the entire process of "perception-analysis-early warning-handling-optimization" for safety incidents in the teaching environment. It comprehensively covers the safety protection needs of all teaching scenarios such as teaching buildings, laboratories, and dormitories, effectively protecting the personal safety of teachers and students, the safety of campus property, and maintaining normal campus teaching order. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the method for ensuring the effectiveness of this invention.

[0018] Figure 2 This is a diagram of the three-layer edge computing node architecture of the present invention. Detailed Implementation

[0019] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0020] Example according to Figure 1 and Figure 2 As shown, this embodiment provides a teaching environment security method based on edge computing, implemented using a three-layer edge computing node architecture consisting of regional edge nodes, campus edge center nodes, and cloud collaborative nodes. Each regional edge node communicates with the campus edge center node via 5G / wired LAN for low-latency communication, uploading only abnormal feature data, early warning information, and handling results to the campus edge center node, avoiding long-distance transmission of the full original data. The method includes the following steps: Step 1: Divide the teaching area into protected zones according to the security level, and deploy area edge nodes in each zone that integrate multi-source sensing devices and lightweight analysis modules with built-in lightweight analysis models. Regional edge nodes collect and preprocess local multi-source sensing data for subsequent analysis. Core data involving privacy are stored locally only (such as dormitory interior images and classroom personnel information) and are not uploaded to the cloud. Abnormal feature data is uploaded to the campus edge central node. Data preprocessing includes format conversion, invalid data removal, and redundant data compression.

[0021] Multi-source sensing devices include one or more of the following: high-definition cameras (with human recognition and behavior analysis functions), environmental sensors (smoke detectors, temperature sensors, gas / hazardous chemical detectors, air quality sensors), equipment status sensors (laboratory instrument operation sensors, classroom electrical appliances sensors, dormitory electrical safety sensors), personnel density sensors, and access control sensing terminals (facial recognition / card swiping).

[0022] The protection zone is specifically divided into a core protection zone, a general protection zone, and an outer protection zone. The core protection zone includes laboratories, student dormitories, hazardous chemical warehouses, and training centers. The general protection zone includes classrooms in teaching buildings, libraries, and canteens. The outer protection zone includes campus corridors, playgrounds, and recreational areas.

[0023] The lightweight analysis model supports local offline operation, enabling local data analysis and coordinated device handling even in the event of network interruption; the campus edge center node has node redundancy backup function to avoid cross-regional scheduling failure caused by a single device failure.

[0024] Step 2: Perform real-time analysis on the pre-processed data locally using a lightweight analysis model within the regional edge nodes. Combine this with a pre-built knowledge base of safety features for teaching scenarios to identify safety events and label them with regional warning levels. Then, perform cross-regional correlation analysis on the abnormal feature data uploaded from each region through the campus edge center node to identify composite risks and label them with campus-level warning levels. The campus edge center node is deployed in the campus security management center, serving as the core hub for edge nodes in various regions, enabling the aggregation of data from edge nodes in various regions, cross-regional correlation analysis, and campus-level collaborative scheduling; The knowledge base for safety features in teaching scenarios pre-stores characteristic parameters and judgment thresholds for different teaching areas and different types of safety events, and supports custom settings and updates; Regional early warning levels include Level 1, Level 2, and Level 3. Level 1 warnings correspond to emergency events requiring immediate coordination and response, Level 2 warnings correspond to dangerous events, and Level 3 warnings correspond to general events.

[0025] Step 3: Based on the regional warning level, the edge node of the area that triggered the warning takes the lead in linking the security equipment in the area and issuing disposal instructions to the designated management personnel. Based on the campus warning level, the central node of the campus edge takes the lead in carrying out cross-regional equipment linkage and personnel coordination and dispatch. The response to a Level 1 warning includes the activation of audible and visual alarms by the regional edge nodes, the shutdown of corresponding valves or power supplies, the activation of sprinkler or smoke extraction devices, and the push of warning information containing the location, type, and real-time video of the event (including the location, type, real-time video, and response requirements) to the handheld terminals of regional management personnel (laboratory staff, dormitory staff, and security personnel), requiring them to arrive at the scene within 5 minutes to respond. The response to a Level 2 alert includes coordinating with nearby high-definition cameras to locate the target, sending alert information to security personnel and area administrators, and initiating on-site investigation and interception procedures. The handling process for Level 3 early warnings includes sending an early warning notification only to the area administrator and initiating on-site evacuation / equipment inspection procedures.

[0026] Campus-level early warning and response specifically targets multi-area linkage anomalies and major security incidents (such as simultaneous fires in multiple areas of the campus or large-scale intrusions by off-campus personnel). The campus perimeter central node coordinates and dispatches perimeter nodes in various areas to achieve cross-area equipment linkage (such as linkage of cameras in key areas of the entire campus, locking of all access control in the core protection area, and unified activation of sound and light alarms throughout the campus) and personnel collaborative dispatch (such as dispatching campus security teams to form containment / response echelons and pushing the overall situation to the campus security management center in real time), while activating the campus emergency response plan.

[0027] During campus-level early warning and response, each edge node collects response feedback data in real time and synchronizes it to the campus edge center node to achieve dynamic tracking of the response process. The response feedback data includes the arrival time of security personnel, on-site response measures, and the development status of the incident.

[0028] Step 4: The campus edge center node encrypts and uploads the entire process data of the security incident to the cloud collaborative node, which then stores the data, trains the model, and iteratively optimizes the lightweight analysis model and teaching scenario security feature library distributed to the regional edge nodes. The cloud-based collaborative nodes iteratively optimize the lightweight analysis model based on a large database of historical security events. The model is trained using machine learning algorithms, and the optimized model parameters are then distributed to edge nodes in various regions for local updates.

[0029] Step 5: If a regional-level early warning is not handled within the preset time or the incident escalates, it will be automatically reported to the campus edge center node for secondary evaluation, and cloud-based collaborative nodes will be triggered to intervene to expand the scope of handling (such as coordinating with external agencies such as the campus police, fire department, and medical services) until the incident is handled and confirmed, forming a closed-loop management of "perception-analysis-early warning-handling-confirmation".

[0030] When the campus edge central node or the regional edge node fails or loses network connection, the cloud-based collaborative node temporarily takes over the analysis and scheduling tasks to ensure that the system is not interrupted.

[0031] This method employs a hierarchical analysis model combining local lightweight analysis at regional edge nodes and global correlation analysis at campus edge center nodes. It integrates a safety feature database for teaching scenarios to achieve accurate identification and severity rating of safety incidents. The database pre-stores characteristic parameters of typical safety incidents in each region (e.g., laboratory gas leak: gas concentration ≥ 0.5% LEL + smoke sensor not triggered; dormitory electrical fire: sudden current increase ≥ 8A + temperature sensor ≥ 60℃; crowd gathering: density ≥ 30 people / m²). 2 It can stay for ≥3 minutes and supports local custom updates.

[0032] Regional edge node analysis involves deploying corresponding lightweight analysis models (such as hazardous chemical testing and equipment operation specification analysis models in laboratories, electrical safety and personnel anomaly analysis models in dormitories, and personnel fall and gathering analysis models in classrooms) to address the safety needs of each zone. This allows for real-time analysis of pre-processed local multi-source sensing data, rapid identification of safety events, and labeling of regional warning levels based on the severity of the event (Level 1: Emergency events, such as fires, chemical leaks, and personnel unconsciousness; Level 2: Dangerous events, such as violations of regulations and unauthorized entry by off-campus personnel; Level 3: General events, such as minor personnel gatherings and minor equipment malfunctions).

[0033] The global analysis of the campus edge central node summarizes the abnormal feature data and early warning information uploaded by edge nodes in various areas, constructs a global security situation map of the teaching environment, and conducts cross-regional correlation analysis (such as correlation between data on unauthorized personnel entering the school gate and data on people loitering around the teaching building, and correlation between data on hazardous chemical leaks in laboratories and data on population density in the surrounding area). It identifies cross-regional security risks and marks the campus-level early warning level (only for multi-regional linkage anomalies and major security incidents). During the analysis, if a suspected security incident is identified by an edge node in a region, local data from surrounding sensing devices can be temporarily retrieved for secondary verification to ensure the accuracy of identification.

[0034] Edge-cloud collaboration and global strategy iterative optimization involve the campus edge center node encrypting and uploading full-process data of security incidents (sensing data, analysis results, early warning information, handling process, and handling results) and edge node operational data (device status, communication quality, and analysis accuracy) to the cloud collaboration node. This achieves complementary advantages between the edge and cloud sides, specifically including: Massive data storage and event review: Cloud-based collaborative nodes store full-process data for a long time, building a large database of safety incidents in the teaching environment. This allows campus management departments to retrieve full-process data of any event for review, analysis of safety hazard patterns, and optimization of handling processes. Lightweight analysis model iterative optimization: Based on a large database, the cloud uses machine learning algorithms (such as gradient boosting and neural networks) to train and iterate the lightweight analysis model of the regional edge nodes, optimize the model's feature parameters and recognition accuracy, and then distribute the optimized model to each regional edge node for local updates, thereby improving the recognition accuracy at the edge. Dynamic adjustment of security strategies: Based on the handling data of multiple batches of security incidents, the cloud optimizes the security feature database and hierarchical handling strategies for teaching scenarios (such as adjusting warning thresholds, optimizing equipment linkage rules, and updating personnel scheduling plans), and distributes them to the campus edge center node and regional edge node; Fault protection and cross-campus collaboration: When a campus edge central node or regional edge node fails or loses network access, the cloud-based collaborative node can temporarily take over core analysis and scheduling tasks to ensure that the security system is not interrupted; for multi-campus institutions / education groups, the cloud enables data communication and cross-campus collaborative scheduling between edge nodes of each campus, realizing the overall security coordination of the entire teaching environment.

[0035] This method can also access data from the campus smart teaching platform, combining teaching schedules and student and teacher attendance data to achieve more accurate safety analysis (such as monitoring electrical safety in empty classrooms and monitoring the use of hazardous chemicals in laboratories during experimental classes); it also supports visual management via mobile APP, allowing campus administrators to view the safety status of each area in real time, receive early warning information, and issue remote handling instructions via mobile phone / tablet.

[0036] Application examples This embodiment uses the security of the teaching environment of an undergraduate university as an application scenario to specifically explain the proposed edge computing-based method for ensuring the security of the teaching environment.

[0037] In this embodiment, all three edge computing nodes adopt industrial-grade hardware configurations. The lightweight analysis model of the regional edge nodes is a lightweight model based on YOLOv8 compression. The campus network adopts a dual-network convergence mode of 5G + gigabit wired LAN to ensure low latency and stability of data transmission.

[0038] Step 1: Deployment of three-layer edge nodes and localized collection of multi-source sensing data Teaching area division: The school's teaching environment is divided into a core protection zone (2 chemistry laboratories, 1 biology laboratory, 3 student dormitories, 1 hazardous chemical warehouse), a general protection zone (8 teaching buildings, 2 libraries, 3 canteens), and an outer protection zone (passages around the teaching buildings, 2 playgrounds, and campus recreational areas).

[0039] Node Deployment: Twelve regional edge nodes are deployed in each zone. Each node is configured with an edge computing gateway, a lightweight YOLOv8 analysis module, and corresponding sensing devices (gas / hazardous chemical sensors, electricity sensors, and high-definition cameras are deployed in the chemistry lab; electricity safety sensors, temperature / smoke sensors, and facial recognition access control are deployed in student dormitories; personnel density sensors, cameras, and classroom electrical status sensors are deployed in the teaching building); one campus edge center node is deployed in the campus security management center, configured with four edge servers, a global situation analysis platform, and a collaborative scheduling module; and one cloud collaborative node is deployed on the school's private cloud platform, configured with a big data storage server, a model training and optimization module, and a cross-campus collaborative module.

[0040] Data Acquisition: Multi-source sensing devices at the edge nodes of each area collect data at a preset frequency (real-time acquisition by laboratory sensors, 1 frame / second from cameras, and 5 times / second from dormitory power consumption sensors). After acquisition, local preprocessing is performed immediately (removing blurry video frames, filtering sensor drift data, and compressing video data). Only abnormal characteristic data (such as excessive gas concentration and sudden current increase) are uploaded to the campus edge center node. Privacy data such as dormitory internal images and classroom personnel information are stored only at the local area edge node.

[0041] Step 2: Layered Edge Analysis and Precise Identification of Security Incidents Local analysis of regional edge nodes: Each regional edge node performs real-time analysis of preprocessed data through a corresponding lightweight analysis model, and identifies security events by combining the teaching scenario security feature library.

[0042] For example, if a gas sensor at the edge node of a chemical laboratory detects a gas leak rate of 0.6% LEL (exceeding the warning threshold of 0.5% LEL), and a high-definition camera captures data indicating that personnel have not evacuated from the laboratory, the analysis using a lightweight hazardous chemical detection model will determine it as a regional level 1 warning, and the event type will be labeled as "laboratory gas leak".

[0043] For example, the edge node of the teaching building detected a population density of 35 people / m² in a certain corridor. 2 The incident occurred within 4 minutes, triggering a Level 3 regional alert, with the event type labeled as "minor gathering of people in the corridor".

[0044] Global Analysis of Campus Edge Center Nodes: The campus edge center nodes aggregate early warning information and abnormal feature data from edge nodes in various areas to construct a global security situation map. Suppose that at a certain moment, the edge node in the school gate area uploads level-two early warning data of "external personnel failing facial recognition and forcibly breaking in," and the edge node in the playground area uploads feature data of "the external personnel abnormally loitering on the playground." The campus edge center nodes, through cross-regional correlation analysis, determine that it is a campus-level level-two early warning, and the event type is labeled as "external personnel dangerously wandering across regions."

[0045] Step 3: Tiered early warning and real-time response at the edge Level 1 Regional Warning Response (Laboratory Gas Leak): The local linkage response will be immediately triggered at the edge node of the chemical laboratory: shut off the main gas valve and main power switch of the laboratory, start the sprinkler system and smoke exhaust system of the laboratory, turn on the audible and visual alarms in the laboratory and surrounding corridors, and lock the access control of the laboratory and adjacent areas; at the same time, a warning message will be pushed to the handheld terminals of the laboratory technicians and campus security personnel (the 3 people closest to the laboratory), including "Gas leak in chemical laboratory 302, Level 1 warning, please arrive at the scene within 5 minutes to handle the situation, the linkage protection has been activated on site" and real-time video of the laboratory.

[0046] Campus-level Level 2 Early Warning Response (Dangerous Loitering of Off-Campus Personnel Across Areas): The central node at the perimeter of the campus immediately initiates cross-regional collaborative response: all cameras around the school gates, playground, and teaching buildings are deployed to lock onto the off-campus personnel, enabling real-time tracking of their trajectory; the personnel's real-time trajectory and location information are pushed to the dispatch terminal of the campus security team, and security personnel are dispatched to form an interception echelon at the passageway between the playground and the teaching building; at the same time, overall situational information is pushed to the on-duty personnel of the campus security management center to achieve real-time monitoring and command.

[0047] Regional Level 3 Early Warning Response (Slight Gathering of People in Corridors): The regional edge node of the teaching building only pushes an early warning message to the administrator of that teaching building: "Slight gathering of people in the corridor on the 2nd floor of Building 3. Please go and guide them." After the administrator arrives at the scene and completes the guidance, the response is confirmed to be completed on the terminal.

[0048] Step 4: Edge-Cloud Collaboration and Global Strategy Iterative Optimization The campus edge central node encrypted and uploaded all data (sensing data, analysis results, early warning information, handling process, and handling results) of the "laboratory gas leak" and "dangerous cross-regional wandering of off-campus personnel" to the cloud collaborative node: Data storage and review: Data is stored in the cloud into a big data database of safety incidents in the teaching environment. The campus safety management department can retrieve the full process data of the incident through the cloud platform, review the handling process (such as the arrival time of security personnel being 3 minutes, which meets the handling requirements; the response time of laboratory linkage equipment being 0.3 seconds), and analyze optimization points.

[0049] Model iteration and optimization: Based on data from 120 safety incidents in the past 6 months in the big data database, the cloud trained a lightweight YOLOv8 analysis model for the regional edge nodes, optimized the feature recognition parameters for gas leaks and personnel intrusions, and distributed the optimized model to each regional edge node for local updates. After the update, the model's recognition accuracy improved from 92% to 98.5%.

[0050] Strategy Adjustment: Based on the handling data of unauthorized personnel entering the campus, the cloud optimizes the feature parameters of "dangerous unauthorized personnel wandering around" in the teaching scenario security feature database (adding the judgment condition of "crossing two or more areas within 10 minutes after facial recognition failure"), and distributes the optimized feature database to the campus edge center node and the edge nodes of each area.

[0051] Step 5: Early Warning Escalation and Closed-Loop Response Mechanism Suppose that in a certain scenario, the edge node of a student dormitory area detects a sudden increase in electrical current to 9A, triggering a level-one regional warning (risk of electrical fire in the dormitory), and pushes disposal instructions to dormitory management and security personnel. However, due to an urgent task, the security personnel have not arrived at the scene within 8 minutes, and the temperature sensor detects that the temperature has risen to 70°C, indicating that the risk of the incident is escalating. At this point, the dormitory area edge node automatically reports the warning level to the campus edge center node. After a secondary assessment by the campus edge center node, the cloud-based collaborative node is triggered to intervene. The cloud immediately links with the school fire department and school clinic, dispatching firefighters and medical personnel to the dormitory area and activating the campus emergency broadcast to achieve the highest level of emergency response. Until the firefighters arrive at the scene and eliminate the fire risk, and the disposal personnel confirm the completion of the disposal on the terminal, this safety incident forms a closed-loop management system.

[0052] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for ensuring the security of a teaching environment based on edge computing, characterized in that, The three-layer edge computing node architecture, consisting of regional edge nodes, campus edge center nodes, and cloud collaboration nodes, is implemented through the following steps: Step 1: Divide the teaching area into protected zones according to the security level, and deploy edge nodes in each zone; Step 2: Perform real-time analysis of the data using a lightweight analysis model, then identify security events and label them with regional warning levels. Through the campus edge center node, perform cross-regional correlation analysis on abnormal feature data, identify composite risks, and label them with campus-level warning levels. Step 3: Implement early warning measures based on the regional and campus-level early warning levels; Step 4: The campus edge center node encrypts and uploads the entire process data of the security incident to the cloud collaborative node for data storage and model training, and iteratively optimizes the lightweight analysis model and the teaching scenario security feature library. Step 5: Secondary assessment of regional early warnings and handling of escalated incidents to form a closed-loop management system.

2. The method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: In step one, the regional edge node integrates multi-source sensing devices and a lightweight analysis module with a built-in lightweight analysis model to collect and preprocess local multi-source sensing data for subsequent analysis. Core data involving privacy is stored locally only, while abnormal feature data is uploaded to the campus edge center node. Data preprocessing includes format conversion, invalid data removal, and redundant data compression.

3. The method for ensuring the security of a teaching environment based on edge computing according to claim 2, characterized in that: The multi-source sensing device includes one or more of the following: high-definition camera, environmental sensor, equipment status sensor, personnel density sensor, and access control sensing terminal.

4. The method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: The division of the protection zone in step one specifically involves dividing the teaching environment into a core protection zone, a general protection zone, and an outer protection zone. The core protection zone includes laboratories, student dormitories, hazardous chemical warehouses, and training centers. The general protection zone includes classrooms in teaching buildings, libraries, and canteens. The outer protection zone includes campus corridors, playgrounds, and recreational areas.

5. The method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: In step two, the identification of safety events specifically involves using a pre-set knowledge base of safety characteristics for teaching scenarios. This knowledge base stores characteristic parameters and judgment thresholds for different teaching areas and different types of safety events, and supports custom settings and updates. The regional warning levels include Level 1, Level 2, and Level 3 warnings. Level 1 warnings correspond to emergency events requiring immediate action, Level 2 warnings correspond to dangerous events, and Level 3 warnings correspond to general events.

6. The method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: The specific early warning response in step three is as follows: based on the regional early warning level, the edge node of the area that triggered the early warning takes the lead, links the security equipment in the area, and issues response instructions to the designated management personnel; based on the campus early warning level, the central node at the campus edge takes the lead, and conducts cross-regional equipment linkage and personnel coordination and dispatch.

7. A method for ensuring the security of a teaching environment based on edge computing according to claim 6, characterized in that: When handling the early warning, the handling of a Level 1 early warning includes activating the audible and visual alarms by linking the regional edge nodes, shutting off the corresponding valves or power supplies, starting the sprinkler or smoke exhaust devices, and pushing early warning information containing the event location, type, and real-time video to the handheld terminals of regional management personnel. The response to a Level 2 alert includes coordinating with nearby high-definition cameras to locate the target, sending alert information to security personnel and area administrators, and initiating on-site investigation and interception procedures. The handling process for Level 3 early warnings includes sending an early warning notification only to the area administrator and initiating on-site evacuation / equipment inspection procedures.

8. A method for ensuring the security of a teaching environment based on edge computing according to claim 6, characterized in that: During the campus-level early warning and response process, each edge node collects response feedback data in real time and synchronizes it to the campus edge center node to achieve dynamic tracking of the response process. The response feedback data includes the arrival time of security personnel, on-site response measures, and the development status of the incident.

9. A method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: In step four, the cloud-based collaborative nodes iteratively optimize the lightweight analysis model based on a large database of historical security events. The model is trained using machine learning algorithms, and the optimized model parameters are then distributed to edge nodes in each region for local updates.

10. A method for ensuring the security of a teaching environment based on edge computing according to claim 1, characterized in that: Specifically, step five involves automatically reporting to the campus edge center node for secondary evaluation if the regional-level early warning is not handled within a preset time or if the event escalates, and triggering the intervention of cloud-based collaborative nodes to expand the scope of handling until the event is handled and confirmed. It also includes the provision that when the campus edge center node or the regional edge node fails or loses network access, the cloud collaborative node temporarily takes over the analysis and scheduling tasks to ensure that the system is not interrupted.

Images