A globalized security management system and method

Abstract

The application relates to the technical field of engineering construction safety management, and discloses a globalized safety management system and method, which aims to solve the problems of insufficient dynamic adaptation of traditional safety management, lagging hidden danger identification, difficult responsibility tracing and low-efficiency emergency response. The system comprises a five-level architecture of a sensing layer, a network layer, a data center layer, an application layer and a display layer, and the application layer is integrated with a dynamic self-adaptive grid management module, an AI multi-source data identification and early warning module, a blockchain handover and tracing module, a data center prediction module and an intelligent emergency decision module. The method constructs a whole-period, whole-space and whole-process management and control system through dynamic adaptive grids, real-time hidden danger identification, traceable handover, predictive prevention and control and intelligent emergency response. The application realizes dynamic adaptation, accurate identification, traceable responsibility, active prevention and control and intelligent emergency response of safety management, greatly improves the lean and intelligent level of engineering construction safety management, and guarantees safety production in the whole cycle of a project.

Description

Technical Field

[0001] This invention relates to the field of engineering construction safety management technology, and in particular to a comprehensive safety management system and method. Background Technology

[0002] As engineering construction projects continue to expand in scale and construction environments become increasingly complex, the characteristics of dispersed high-risk work sites, dynamic changes in construction phases, and continuous work time are becoming more prominent. This places higher demands on the comprehensiveness, accuracy, and timeliness of on-site safety management. Although the current safety management model in the engineering construction field has gradually introduced methods such as grid management, drone inspections, and standardized shift handover, there are still many technical pain points that are difficult to solve in practical applications, which cannot meet the needs of comprehensive and intelligent safety control.

[0003] In terms of regional control, traditional safety management often adopts a static grid-based approach. Once the boundaries, scope, and personnel configuration of the grid are determined, they remain fixed for a long period. This makes it difficult to adapt to the actual scenarios of construction progress, dynamic changes in risk levels, and adjustments in the distribution of personnel and equipment. This static management model can easily lead to insufficient supervision in high-risk areas during peak construction periods, duplicate supervision in low-risk areas, or blurred responsibility boundaries and supervision vacuums due to adjustments in construction procedures. At the same time, the mismatch between the personnel monitoring the grid risk level and operational needs seriously affects the coverage quality and execution efficiency of safety supervision.

[0004] In terms of hazard identification and accountability, traditional safety management relies mainly on manual inspections and simple equipment. This not only makes it difficult to cover special areas such as high places and corners, but also relies on the experience and judgment of inspection personnel for hazard identification, which can lead to problems such as delayed identification, missed identification, and misjudgment. At the same time, the handover process lacks a standardized intelligent verification mechanism. Relying solely on manual recording of handover content can easily result in the omission of key information and unclear definition of responsibility. Furthermore, the handover data lacks tamper-proof storage methods, making it difficult to trace responsibility in the event of a safety accident and failing to effectively avoid the risk of a gap in accountability.

[0005] In terms of risk prevention and emergency response, the existing safety management model is mostly based on post-event response and lacks the ability to proactively predict potential risks. Due to the lack of a multi-dimensional data correlation analysis and intelligent prediction mechanism, it is impossible to identify high-risk areas and potential hazard types in advance, resulting in a lack of targeted prevention and control measures and difficulty in reducing the probability of accidents from the source. In addition, during the emergency response process, the formulation of rescue plans relies on manual decision-making, which is greatly limited by experience, and the dispatch of rescue resources lacks intelligent support, making it impossible to quickly match the optimal rescue path and resource allocation, resulting in low emergency response efficiency and difficulty in effectively reducing accident losses.

[0006] Therefore, this invention proposes a comprehensive security management system and method. Summary of the Invention

[0007] One objective of this invention is to propose a comprehensive safety management system and method. This invention can achieve dynamic adaptation, real-time identification, traceable handover, predictive prevention and control, and intelligent emergency response in construction safety management. It completely solves the pain points in traditional safety management, such as regulatory vacuum, repetitive supervision, responsibility gaps, delayed hazard identification, and inefficient emergency response, and builds an integrated safety management and control system that covers all time periods, all spaces, and all processes.

[0008] According to an embodiment of the present invention, a comprehensive security management system includes a perception layer, a network layer, a data platform layer, an application layer, and a presentation layer.

[0009] The application layer integrates a dynamic adaptive grid management module, an AI-fusion multi-source data real-time risk identification and early warning module, an intelligent handover verification and responsibility tracing module, a full-domain safety data platform module, and an emergency intelligent decision support module. These modules work together to achieve dynamic adaptation, real-time identification, traceable handover, predictive prevention and control, and intelligent emergency response in construction safety management.

[0010] The dynamic adaptive grid management module, based on three-dimensional data of construction progress, real-time risk level, and personnel and equipment distribution, automatically adjusts grid division, boundaries, and monitoring personnel configuration through algorithms. The AI-integrated multi-source data real-time risk identification and early warning module integrates multi-source data from the perception layer and achieves real-time identification and graded early warning of hidden dangers through a multi-modal AI model. The intelligent handover verification and responsibility traceability module combines blockchain and AI technologies to achieve intelligent verification of handover content, solidification of responsibilities, and full-process traceability. The full-domain safety data platform module integrates multi-dimensional management data and achieves risk prediction through correlation analysis and machine learning. The emergency intelligent decision support module automatically generates the optimal rescue plan and dispatches resources based on real-time scene data.

[0011] Furthermore, the dynamic adaptive grid management module includes a grid partitioning algorithm and a grid dynamic update mechanism. The grid partitioning algorithm takes into account the construction stage, risk level, personnel density, and equipment concentration data, and calculates the optimal grid partitioning scheme through a multi-objective optimization function. The grid dynamic update mechanism collects real-time data at a preset cycle, automatically judges the grid adjustment needs and adjusts the results synchronously, and triggers personnel scheduling reminders to achieve dynamic matching of responsibilities and grids.

[0012] Furthermore, the AI-integrated multi-source data real-time risk identification and early warning module integrates drone inspection data, mobile surveillance video data, and multi-dimensional sensor data. The multi-dimensional sensors include power sensors, fire sensors, and environmental sensors. The multi-modal AI model includes target detection algorithms and convolutional neural networks, which are used to identify violations, equipment malfunctions, and environmental exceedance risks, respectively, to achieve graded early warning push.

[0013] Furthermore, the intelligent handover verification and responsibility traceability module includes an intelligent handover content input unit, an AI handover verification unit, a blockchain responsibility solidification unit, and an intelligent responsibility sign terminal. The AI ​​handover verification unit compares historical hazard data with real-time early warning data to verify the completeness and accuracy of the handover content. The blockchain responsibility solidification unit stores the verified handover data on the blockchain and generates a unique identifier to ensure that the data cannot be tampered with.

[0014] Furthermore, the comprehensive security data platform module includes a data fusion unit, a risk prediction model, and a hybrid storage architecture. The data fusion unit uses ontology modeling technology to establish a correlation model of personnel, equipment, environment, and management. The risk prediction model is based on a temporal neural network, which takes into account historical hazard data, equipment operation data, personnel training records, and environmental data to predict high-risk areas and potential hazard types within a preset time period. The hybrid storage architecture uses blockchain and a distributed database to store key responsibility data and massive amounts of perception data, respectively.

[0015] Furthermore, the emergency intelligent decision support module includes an accident scenario modeling unit, a rescue plan generation unit, a resource intelligent scheduling unit, and an emergency process tracing unit. The accident scenario modeling unit combines UAV aerial photography data, personnel positioning data, and equipment distribution data to construct a three-dimensional model of the accident site. The rescue plan generation unit uses optimization algorithms to generate rescue routes and resource allocation plans with the goal of minimizing rescue time and reducing casualty risk. The resource intelligent scheduling unit is linked to the rescue resource database, automatically pushes scheduling instructions, and tracks the arrival of resources.

[0016] Furthermore, the network layer includes a 5G communication module, a WiFi 6 module, and a LoRa IoT gateway to achieve high-speed, low-latency transmission of data from the perception layer. The display layer includes a large-scale monitoring screen, a mobile APP terminal, and a smart responsibility sign terminal to display grid information, early warning information, handover records, and emergency plans.

[0017] A comprehensive security management method includes the following steps:

[0018] S1. Data Acquisition: The perception layer devices continuously collect data on personnel, equipment, environment, and construction behavior in the construction scenario, and transmit it to the data platform layer in real time through the network layer.

[0019] S2. Dynamic Grid Adjustment: After the data platform layer processes the data, the dynamic adaptive grid management module automatically adjusts the grid division and the configuration of responsible personnel through algorithms and synchronizes them to each terminal.

[0020] S3. Risk Identification and Early Warning: The AI ​​analysis engine integrates multi-source data to identify potential hazards in real time and pushes early warning information to the corresponding responsible personnel in a tiered manner.

[0021] S4. Intelligent handover: During shift handover, the responsible personnel enter the handover content through the terminal. After AI verification, the information is stored on the blockchain to complete the handover of responsibilities.

[0022] S5. Risk Prediction and Prevention: The full-domain security data platform module uses a risk prediction model to provide early warnings of potential hazards, enabling responsible personnel to take targeted preventive measures.

[0023] S6. Emergency Response: In the event of an accident, the emergency intelligent decision support module automatically generates a rescue plan, allocates resources, tracks the progress of the response, and records the entire process.

[0024] Furthermore, in step S1, the perception layer device includes a drone inspection unit, a mobile monitoring terminal, multi-dimensional sensors, personnel positioning tags, and an equipment operation monitoring module. The data collection scope covers the safety behavior of construction personnel, equipment operating status, environmental parameters, and the execution of construction procedures.

[0025] Furthermore, in step S5, risk prediction and prevention includes the system automatically pushing a prevention task list to the corresponding grid leader and safety supervisor based on the high-risk areas and potential hazard types output by the risk prediction model. The list includes hazard prevention and control measures, inspection frequency and responsibility requirements. After the prevention and control is completed, the prevention and control effect is verified through the perception layer data.

[0026] The beneficial effects of this invention are:

[0027] 1. This invention uses a dynamic adaptive grid management mechanism to achieve real-time adaptation of grid division, boundaries, and monitoring personnel configuration with construction progress, risk level, and personnel and equipment distribution. This completely solves the problems of regulatory vacuum, repetitive supervision, and imbalance in responsibility matching in the traditional model, ensuring comprehensive safety supervision without blind spots and accurate implementation of responsibilities, and greatly improving the dynamic adaptability and coverage quality of safety management.

[0028] 2. This invention integrates multi-dimensional perception data through a multi-modal AI model to achieve real-time and accurate identification and graded early warning of potential hazards. This solves the pain points of limited coverage and delayed identification in traditional manual inspections. Furthermore, it uses AI to intelligently verify the completeness and accuracy of handover content and combines blockchain to ensure that handover data is tamper-proof and traceable, completely eliminating problems such as gaps in responsibility and omissions in content during the handover process. This significantly improves the accuracy of safety management and the reliability of responsibility traceability.

[0029] 3. This invention utilizes the correlation analysis and machine learning capabilities of the full-domain security data platform to predict potential high-risk areas and types of hidden dangers in advance, providing forward-looking support for the formulation of prevention and control measures. At the same time, in emergency scenarios, it relies on real-time scenario modeling and optimization algorithms to automatically generate the optimal rescue plan and intelligently allocate resources, solving the problem of traditional emergency response relying on manual decision-making and low efficiency, and greatly improving the initiative of risk prevention and control and the scientific and efficient nature of emergency response. Attached Figure Description

[0030] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0031] Figure 1 This is a schematic diagram of the overall framework of a comprehensive security management system proposed in this invention;

[0032] Figure 2 This is a flowchart illustrating the operation of a comprehensive security management method proposed in this invention. Detailed Implementation

[0033] To make the technical means and objectives and effects of the present invention easier to understand, the embodiments of the present invention will be described in detail below with reference to specific illustrations.

[0034] Example 1

[0035] like Figure 1 As shown, this invention discloses a global security management system, including a perception layer, a network layer, a data platform layer, an application layer, and a presentation layer;

[0036] The perception layer, as a data acquisition terminal, needs to cover all work areas and key locations throughout the construction site, including drone inspection units, mobile monitoring terminals, multi-dimensional sensors, personnel positioning tags, and equipment operation monitoring modules. The drone inspection unit is equipped with a high-definition image acquisition module and an infrared thermal imaging module, and performs aerial inspection tasks according to a preset cruise route or remote control commands. The mobile monitoring terminal is installed on a movable bracket and deployed in the work area of ​​the construction team and temporary work points to achieve close-range video acquisition. The multi-dimensional sensors include power sensors, fire sensors and environmental sensors. The power sensors are connected in series in the construction distribution box and the power supply circuit of electrical equipment to collect electrical parameters such as voltage, current and leakage current in real time. The fire sensors are installed at fire extinguisher storage points, fire hydrants and flammable and explosive material storage areas to collect parameters such as fire extinguisher pressure, fire hydrant water level and combustible gas concentration. The environmental sensors are evenly distributed on the construction site to collect environmental parameters such as temperature, humidity, wind speed and dust concentration. The personnel positioning tag adopts a wearable design and is carried by the construction personnel. It has a built-in positioning chip and communication module to provide real-time feedback of personnel location information. The equipment operation monitoring module is installed on large construction machinery such as tower cranes, excavators and cranes to collect operating parameters such as equipment speed, load and vibration frequency. All sensing layer devices have real-time data output function to ensure the timeliness of the collected data.

[0037] The network layer comprises a 5G communication module, a WiFi 6 module, and a LoRa IoT gateway. These three components work together to achieve high-speed, low-latency transmission of data from the perception layer. The 5G communication module is used for long-distance transmission of large-capacity data such as drone inspection data and high-definition video data. Leveraging the high bandwidth and low latency of the 5G network, it ensures a data transmission rate of no less than 100Mbps and a transmission latency of no more than 20ms. The WiFi 6 module is deployed in construction living areas, office areas, and core operational areas, covering short-range data transmission needs. It supports simultaneous access for multiple devices, with a maximum of 64 connected devices and a single device transmission rate of no less than 50Mbps. The LoRa IoT gateway is used for data transmission from low-power devices such as multi-dimensional sensors and personnel positioning tags. Utilizing LoRa spread spectrum communication technology, it achieves a communication distance of 1-3km and supports star and mesh networking methods, ensuring stable data transmission from edge devices. All communication modules in the network layer aggregate data through routing devices and transmit it uniformly to the data platform layer.

[0038] The data platform layer comprises blockchain storage nodes, an AI analysis engine, a data fusion module, and a risk prediction model, undertaking data storage, processing, fusion analysis, and risk prediction functions. The blockchain storage nodes adopt a Hyperledger Fabric consortium blockchain architecture, deploying five consensus nodes, maintained by the project management, construction, supervision, safety supervision, and technical support parties respectively, ensuring decentralized and secure data storage. The AI ​​analysis engine employs a GPU cluster architecture, configured with multiple high-performance computing graphics cards, providing sufficient computing power to support the operation of multimodal AI models. The data fusion module uses an ETL (Extract-Transform-Load) data processing flow, first extracting the raw data collected by various devices in the perception layer, then converting the heterogeneous data into a unified format through data cleaning, format conversion, deduplication, and noise reduction operations, and finally loading it into the data storage unit. The risk prediction model is built based on a temporal neural network to achieve predictive analysis of potential risks.

[0039] The application layer integrates a dynamic adaptive grid management module, an AI-integrated multi-source data real-time risk identification and early warning module, an intelligent handover verification and responsibility tracing module, a full-domain security data platform module, and an emergency intelligent decision support module. These modules work together to achieve core business functions.

[0040] The dynamic adaptive grid management module includes a grid partitioning algorithm and a dynamic grid update mechanism. The grid partitioning algorithm takes construction stage, risk level, personnel density, and equipment concentration data as input, and calculates the optimal grid partitioning scheme through a multi-objective optimization function. The expression of the multi-objective optimization function is as follows:

[0041]

[0042]

[0043]

[0044] in, For the decision variable vector, Indicates the number of grid cells. Represents the area of ​​a single grid cell. Represents the coordinates of the grid boundary. This indicates the number of monitoring personnel within the grid; the rest... Determine the relevant auxiliary variables for the grid;

[0045] This is a function for the area of ​​the regulatory blind zone, used to minimize the grid coverage blind zone. , The total area of ​​the construction site. The effective coverage area of ​​the k-th grid;

[0046] For the workload balancing function of monitoring personnel, , Let k be the number of construction workers in the k-th grid. Let k be the number of devices in the k-th grid. This represents the number of monitoring personnel within the k-th grid.

[0047] For the risk coverage completeness function, , The risk level coefficient for the k-th grid is... This refers to the average risk level coefficient at the construction site.

[0048] , Decision variables The lower and upper limits, such as the area of ​​a single grid cell. lower limit 50 square meters, maximum value It is 500 square meters;

[0049] These are constraint functions, including non-overlapping grid boundaries, matching of monitoring personnel qualifications, and equipment coverage radius constraints.

[0050] The grid dynamic update mechanism collects real-time data transmitted from the perception layer at a preset cycle, recalculates the optimal grid partitioning scheme through the aforementioned multi-objective optimization function, compares it with the current grid partitioning scheme, and automatically triggers grid adjustment if the scheme difference exceeds a preset threshold. The adjustment result is synchronized to each terminal in the display layer through the network layer and a scheduling reminder message is pushed to relevant monitoring personnel to achieve dynamic matching of responsibility and grid.

[0051] The AI-integrated multi-source data real-time risk identification and early warning module integrates drone inspection data, mobile surveillance video data, and multi-dimensional sensor data to achieve real-time identification and graded early warning of potential hazards through a multi-modal AI model.

[0052] The multimodal AI model includes an object detection algorithm and a convolutional neural network. The object detection algorithm uses the YOLOv8 model to identify violations such as not wearing a safety helmet, unauthorized hot work, and not wearing a safety belt while working at height. Its loss function is the CIoU loss function, expressed as follows:

[0053]

[0054]

[0055]

[0056]

[0057] in, To predict the intersection-union ratio (IoU) between the bounding box and the ground truth bounding box, To predict the center coordinates of the bounding box, The center coordinates of the true bounding box. for and The square of the Euclidean distance between them The length of the diagonal of the smallest bounding rectangle that encloses the predicted bounding box and the true bounding box. , To predict the width and height of the bounding box, , The width and height of the actual bounding box. This is a parameter for maintaining the aspect ratio consistency of the bounding box. This is the balance coefficient.

[0058] Convolutional neural networks are used to identify equipment malfunctions and environmental risks exceeding standards. The calculation expression for their convolutional layers is as follows:

[0059]

[0060] in, For the first Output feature maps of convolutional layers The ReLU function is used as the activation function. For the first The number of feature maps output by the layer. For the first Layer The weight parameters of each convolutional kernel, For the first Layer One input feature map, For the first Layer bias parameters.

[0061] The system calculates the probability of device malfunction and the probability of environmental exceedance by sequentially processing convolutional, pooling, and fully connected layers, and then uses preset thresholds to determine whether to trigger an early warning.

[0062] The AI-integrated multi-source data real-time risk identification and early warning module classifies early warning levels into three levels based on the severity of the hidden dangers: general early warning, major early warning, and critical early warning. General early warnings are pushed to safety monitoring personnel within the grid, major early warnings are pushed to the grid leader and safety monitoring personnel, and critical early warnings are simultaneously pushed to the project leader, supervisor, and safety regulator, ensuring the accurate delivery of early warning information.

[0063] The intelligent handover verification and accountability tracking module includes an intelligent handover content input unit, an AI handover verification unit, a blockchain-based accountability solidification unit, and an intelligent accountability sign terminal. The intelligent handover content input unit allows responsible personnel to input handover information via a mobile app or intelligent accountability sign terminal, including work content, addressed hazards, unfinished tasks, risk points, and current grid status. The system automatically links to real-time data from the data platform layer, filling in basic fields such as grid information and current warning information, reducing manual data entry workload.

[0064] The AI ​​handover verification unit is used to verify the completeness and accuracy of the handover content. The completeness verification uses a completeness score formula for calculation:

[0065]

[0066] in, For completeness score, This refers to the number of handover items actually entered by the responsible personnel. The system pre-determines the number of mandatory handover items, which include core information such as work content, risk points, unfinished items, current responsible person, and handover time.

[0067] Accuracy verification is calculated using the accuracy score formula:

[0068]

[0069] in, To score for accuracy, To ensure the accuracy of the handover items, the system compares historical hazard data, real-time early warning data, personnel location data, and equipment operation data. For example, if the "no unprocessed hazards" recorded in the handover items is inconsistent with the unprocessed hazard information in the system's real-time early warning data, the item is deemed inaccurate.

[0070] when and If the handover content passes verification, the system will automatically prompt for missing or inaccurate items if the above threshold is not met, and the responsible personnel will supplement and correct them before re-verification.

[0071] The blockchain responsibility solidification unit stores the verified handover data on the blockchain and generates a unique identifier using the SHA-256 hash algorithm. The hash value calculation expression is as follows:

[0072]

[0073] in, This is the hash value of the data being handed over, with a length of 256 bits. The data set for handover includes handover content, personnel information, verification results, handover time, etc. The original data is converted into irreversible hash values ​​through a hash algorithm to ensure that the data cannot be tampered with. At the same time, each blockchain node stores the handover data synchronously to achieve full-process traceability.

[0074] The intelligent responsibility sign terminal uses a touch screen and is installed in a prominent position in each grid area. It displays the current grid information, handover records, real-time hazard warning information, contact information of grid leaders and safety supervisors, etc. Responsible personnel can log in by scanning a code or entering a password to carry out handover operations, and the data is synchronized to the data platform layer in real time.

[0075] The comprehensive security data platform module includes a data fusion unit, a risk prediction model, and a hybrid storage architecture. The data fusion unit uses ontology modeling technology to establish a relationship model between personnel, equipment, environment, and management. The ontology model describes the relationship between each element through triples, such as (construction worker A, operation, tower crane B), (tower crane B, operating status, normal), (high temperature environment, impact, tower crane B load capacity), etc. The relationship model realizes the semantic association of multi-dimensional data.

[0076] The risk prediction model is built on an LSTM temporal neural network to predict high-risk areas and potential hazard types within a preset future time period. The core calculation formula of the LSTM neural network is as follows:

[0077] The formula for calculating the forgetting gate is:

[0078]

[0079] The formula for calculating the input gate is:

[0080]

[0081]

[0082] The cell state update formula is:

[0083]

[0084] The formula for calculating the output gate is:

[0085]

[0086]

[0087] in, Output for the forget gate. For input gate output, For output gate output, It is the sigmoid activation function. The hyperbolic tangent activation function is used. , , , These are the weight matrices for the forget gate, input gate, cell state, and output gate, respectively. , , , These are the corresponding bias vectors. This is the hidden state from the previous moment. The input data at the current moment, This represents the current state of the candidate cells. This represents the current state of the cell. To determine the current output state, a fully connected layer is used. This is mapped to risk prediction results.

[0088] The hybrid storage architecture employs blockchain and a distributed database. The blockchain stores critical and sensitive data, ensuring that such data is immutable and traceable. The distributed database uses the Hadoop Distributed File System (HDFS) to store massive amounts of sensitive data, supporting high-concurrency read and write operations and massive storage.

[0089] The emergency intelligent decision support module includes an accident scenario modeling unit, a rescue plan generation unit, a resource intelligent scheduling unit, and an emergency process tracing unit. The accident scenario modeling unit combines UAV aerial photography data, personnel location data, and equipment distribution data, and uses a 3D reconstruction algorithm to construct a 3D model of the accident scene. The model identifies key information such as dangerous areas, the location of trapped personnel, rescue channels, and equipment distribution.

[0090] The rescue plan generation unit uses a genetic algorithm to generate rescue routes and resource allocation schemes with the goal of minimizing rescue time and reducing casualty risk. Its objective function expression is as follows:

[0091]

[0092]

[0093]

[0094] in, The objective function value, , For the weighting coefficients, satisfying The weights are adjusted according to the type of accident and the situation on site.

[0095] The total rescue time includes resource allocation time, arrival time of rescue personnel, and rescue implementation time.

[0096] The casualty risk coefficient is calculated based on the extent of the danger zone, the number of trapped people, and the severity of the accident.

[0097] Let j be the actual number of rescue resources dispatched. Let j be the available quantity of the j-th type of rescue resources. For the types and quantities of rescue resources;

[0098] Let m be the path from the m-th node to the n-th node in the rescue route. A set of paths to dangerous areas. , The number of nodes along the rescue route is constrained to ensure that the rescue route does not pass through dangerous areas and that the number of rescue resources dispatched does not exceed the available number.

[0099] The intelligent resource scheduling unit is linked to the project's rescue resource database. Based on the resource allocation requirements generated by the rescue plan, it automatically pushes scheduling instructions to the corresponding rescue personnel. Simultaneously, it tracks the arrival of resources in real time through personnel positioning tags and equipment positioning modules, and feeds back to the emergency command terminal. The emergency process tracing unit records data from the entire emergency response process in real time and automatically generates accident handling reports, providing data support for subsequent accident analysis, liability determination, and management optimization.

[0100] The display layer includes a large-scale monitoring screen, a mobile app terminal, and smart responsibility sign terminals, used to display grid information, early warning information, handover records, and emergency plans. The large-scale monitoring screen, using a video wall design, is installed in the project monitoring center and displays a real-time 3D map of the entire construction site, the status of each grid, real-time early warning information, drone patrol footage, mobile monitoring footage, risk prediction results, etc. The mobile app terminal supports Android and iOS operating systems and has functions such as receiving early warning information, handover operations, reporting hazards, receiving resource scheduling instructions, and viewing emergency plans. The smart responsibility sign terminals are installed in each grid area, displaying the current grid's responsible personnel, handover records, hazard information within the grid, safety precautions, etc., facilitating viewing and supervision by construction personnel.

[0101] Example 2

[0102] Please see Figure 2 This embodiment discloses a comprehensive security management method, based on a comprehensive security management system of Embodiment 1, including the following steps:

[0103] S1. Data Acquisition: The perception layer devices continuously collect comprehensive data on personnel, equipment, environment, and construction behavior in the construction scenario. The collection scope covers the safety behavior of construction personnel, equipment operating status, environmental parameters, and the execution of construction procedures. Among them, the drone inspection unit performs aerial inspections according to a preset patrol cycle, collecting aerial images and infrared thermal imaging data; the mobile monitoring terminal collects video data of the work area in real time; multi-dimensional sensors output monitoring parameters according to a preset collection frequency; personnel positioning tags periodically send location information; and the equipment operation monitoring module collects equipment operation data in real time. All collected data is transmitted in real time to the data platform layer through the 5G communication module, WiFi 6 module, or LoRa IoT gateway of the network layer. During transmission, encryption algorithms are used to encrypt the data to ensure data transmission security.

[0104] S2. Dynamic Grid Adjustment: After receiving data from the perception layer, the data platform layer performs data cleaning, format conversion, deduplication, and noise reduction through the data fusion module to remove invalid and abnormal data. The processed data is then input into the dynamic adaptive grid management module. The dynamic adaptive grid management module calculates the optimal grid division scheme using a multi-objective optimization function. This function aims to minimize regulatory blind spots, balance personnel workload, and maximize risk coverage. It combines data on construction phase, real-time risk level, personnel density, and equipment concentration to solve for the optimal decision variables, obtaining information such as the number of grids, grid boundaries, and the configuration of monitoring personnel within each grid. The grid dynamic update mechanism repeats the above calculation process at a preset cycle (15 minutes), comparing the newly calculated optimal grid division scheme with the current scheme and calculating the difference. When the difference exceeds 10%, grid adjustment is automatically triggered. The adjustment results are synchronized to all terminals through the network layer, and scheduling reminders are pushed to relevant monitoring personnel, clarifying the new grid responsibility areas and work requirements, thus achieving dynamic matching of responsibility and grid.

[0105] S3. Risk Identification and Early Warning: The AI ​​analysis engine in the data platform layer calls upon multimodal AI models and integrates multi-source data transmitted from the perception layer to perform real-time risk identification. Specifically, the YOLOv8 target detection model performs frame analysis on drone inspection images and mobile surveillance videos to identify violations and output relevant information; the convolutional neural network analyzes sensor data and equipment operation data to identify equipment anomalies and environmental exceedance risks and output relevant information. The AI ​​analysis engine classifies warning levels based on the severity of the identification results and pushes warning information to the terminals of the corresponding responsible personnel, reminding them to take timely action.

[0106] S4. Intelligent Handover: During shift handover, the outgoing shift personnel log in to the intelligent handover verification and accountability module via their terminal, enter handover information, and the system automatically populates the basic fields by referencing the data from the data platform layer. After the data entry is complete, the AI ​​handover verification unit calculates the completeness and accuracy of the handover content using completeness and accuracy scoring formulas. and If the above threshold is not met, the verification is successful; if it is not met, the system will automatically prompt for correction, and verification will be carried out again after the correction is made. After the verification is successful, the blockchain responsibility solidification unit will store the handover data on the blockchain and generate a unique identifier. Each blockchain node will store the data synchronously to complete the handover of responsibility. After the successor confirms receipt, the handover process ends.

[0107] S5. Risk Prediction and Prevention: The risk prediction model of the full-domain security data platform module is based on an LSTM temporal neural network. It takes multi-dimensional input data and calculates the results sequentially through a forget gate, input gate, cell state update, and output gate, outputting predictions of high-risk areas and potential hazard types for the next 24 hours. Based on the prediction results, the system automatically pushes a prevention task list to the corresponding responsible personnel. The list includes hazard control measures, inspection frequency, and responsibility requirements. The responsible personnel implement the control measures according to the list. After completion, the system verifies the control effect through perception layer data. If the preset standards are not met, an adjusted prevention task list is pushed again until the hazard risk is effectively controlled.

[0108] S6. Emergency Response: When a safety accident occurs at the construction site, the sensing layer devices collect real-time data from the accident site and quickly transmit it to the data platform layer. The emergency intelligent decision support module initiates the emergency response process, the accident scenario modeling unit constructs a 3D model of the accident site, the rescue plan generation unit uses a genetic algorithm to generate the optimal rescue route and resource allocation plan with the goal of minimizing rescue time and reducing casualty risk, the resource intelligent scheduling unit links to the rescue resource database, automatically pushes scheduling instructions, and tracks resource availability in real time, and rescue personnel carry out rescue work according to the emergency plan, while the emergency process traceability unit records the entire process data in real time. After the rescue work is completed, the system automatically generates an accident handling report, analyzes the cause of the accident, the determination of responsibility, and the effectiveness of the emergency response, and simultaneously stores accident-related data in the data platform layer for updating the risk prediction model.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A comprehensive security management system, characterized in that, It includes the perception layer, network layer, data platform layer, application layer, and presentation layer; The application layer integrates a dynamic adaptive grid management module, an AI-fusion multi-source data real-time risk identification and early warning module, an intelligent handover verification and responsibility tracing module, a full-domain safety data platform module, and an emergency intelligent decision support module. These modules work together to achieve dynamic adaptation, real-time identification, traceable handover, predictive prevention and control, and intelligent emergency response in construction safety management. The dynamic adaptive grid management module, based on three-dimensional data of construction progress, real-time risk level, and personnel and equipment distribution, automatically adjusts grid division, boundaries, and monitoring personnel configuration through algorithms. The AI-integrated multi-source data real-time risk identification and early warning module integrates multi-source data from the perception layer and achieves real-time identification and graded early warning of hidden dangers through a multi-modal AI model. The intelligent handover verification and responsibility traceability module combines blockchain and AI technologies to achieve intelligent verification of handover content, solidification of responsibilities, and full-process traceability. The full-domain safety data platform module integrates multi-dimensional management data and achieves risk prediction through correlation analysis and machine learning. The emergency intelligent decision support module automatically generates the optimal rescue plan and dispatches resources based on real-time scene data.

2. The comprehensive security management system according to claim 1, characterized in that, The dynamic adaptive grid management module includes a grid division algorithm and a grid dynamic update mechanism. The grid division algorithm takes into account the construction stage, risk level, personnel density, and equipment concentration data, and calculates the optimal grid division scheme through a multi-objective optimization function. The grid dynamic update mechanism collects real-time data at a preset cycle, automatically judges the grid adjustment needs and adjusts the results synchronously, and triggers personnel scheduling reminders to achieve dynamic matching of responsibilities and grids.

3. The comprehensive security management system according to claim 1, characterized in that, The AI-integrated multi-source data real-time risk identification and early warning module integrates drone inspection data, mobile surveillance video data, and multi-dimensional sensor data. The multi-dimensional sensors include power sensors, fire sensors, and environmental sensors. The multi-modal AI model includes target detection algorithms and convolutional neural networks, which are used to identify violations, equipment malfunctions, and environmental exceedance risks, respectively, to achieve graded early warning push.

4. The comprehensive security management system according to claim 1, characterized in that, The intelligent handover verification and responsibility traceability module includes an intelligent handover content input unit, an AI handover verification unit, a blockchain responsibility solidification unit, and an intelligent responsibility sign terminal. The AI ​​handover verification unit compares historical hazard data with real-time early warning data to verify the completeness and accuracy of the handover content. The blockchain responsibility solidification unit stores the verified handover data on the blockchain and generates a unique identifier to ensure that the data cannot be tampered with.

5. A comprehensive security management system according to claim 1, characterized in that, The comprehensive security data platform module includes a data fusion unit, a risk prediction model, and a hybrid storage architecture. The data fusion unit uses ontology modeling technology to establish a correlation model of personnel, equipment, environment, and management. The risk prediction model is based on a temporal neural network, which takes historical hazard data, equipment operation data, personnel training records, and environmental data as input to predict high-risk areas and potential hazard types within a preset time period. The hybrid storage architecture uses blockchain and a distributed database to store key responsibility data and massive amounts of perception data, respectively.

6. The comprehensive security management system according to claim 1, characterized in that, The emergency intelligent decision support module includes an accident scenario modeling unit, a rescue plan generation unit, a resource intelligent scheduling unit, and an emergency process tracing unit. The accident scenario modeling unit combines UAV aerial photography data, personnel positioning data, and equipment distribution data to construct a three-dimensional model of the accident site. The rescue plan generation unit uses optimization algorithms to generate rescue routes and resource allocation plans with the goal of minimizing rescue time and reducing casualty risk. The resource intelligent scheduling unit is linked to the rescue resource database, automatically pushes scheduling instructions, and tracks the arrival of resources.

7. The comprehensive security management system according to claim 1, characterized in that, The network layer includes a 5G communication module, a WiFi 6 module, and a LoRa IoT gateway to achieve high-speed, low-latency transmission of data from the perception layer. The display layer includes a large-scale monitoring screen, a mobile APP terminal, and a smart responsibility sign terminal to display grid information, early warning information, handover records, and emergency plans.

8. A comprehensive security management method, comprising a comprehensive security management system according to any one of claims 1-7, characterized in that, Includes the following steps: S1. Data Acquisition: The perception layer devices continuously collect data on personnel, equipment, environment, and construction behavior in the construction scenario, and transmit it to the data platform layer in real time through the network layer. S2. Dynamic Grid Adjustment: After the data platform layer processes the data, the dynamic adaptive grid management module automatically adjusts the grid division and the configuration of responsible personnel through algorithms and synchronizes them to each terminal. S3. Risk Identification and Early Warning: The AI ​​analysis engine integrates multi-source data to identify potential hazards in real time and pushes early warning information to the corresponding responsible personnel in a tiered manner. S4. Intelligent handover: During shift handover, the responsible personnel enter the handover content through the terminal. After AI verification, the information is stored on the blockchain to complete the handover of responsibilities. S5. Risk Prediction and Prevention: The full-domain security data platform module uses a risk prediction model to provide early warnings of potential hazards, enabling responsible personnel to take targeted preventive measures. S6. Emergency Response: In the event of an accident, the emergency intelligent decision support module automatically generates a rescue plan, allocates resources, tracks the progress of the response, and records the entire process.

9. The comprehensive security management method according to claim 8, characterized in that, In step S1, the perception layer equipment includes a drone inspection unit, a mobile monitoring terminal, multi-dimensional sensors, personnel positioning tags, and an equipment operation monitoring module. The data collection scope covers the safety behavior of construction personnel, equipment operating status, environmental parameters, and the execution of construction procedures.

10. A comprehensive security management method according to claim 8, characterized in that, In step S5, risk prediction and prevention includes the system automatically pushing a prevention task list to the corresponding grid leader and safety supervisor based on the high-risk areas and potential hazard types output by the risk prediction model. The list includes hazard prevention and control measures, inspection frequency and responsibility requirements. After the prevention and control is completed, the prevention and control effect is verified through the perception layer data.

Images