Intelligent security early warning device and method for industrial park

Through the intelligent security early warning device, combined with intelligent decision-making model and AI vision chip, the problem of unreasonable equipment deployment and lack of coordination mechanisms in traditional security systems is solved, efficient security resource allocation and real-time early warning management in industrial parks are realized, and security guarantees are improved .

CN119007363APending Publication Date: 2024-11-22SHANGHAI YIBANG INTELLIGENT TECH CO LTD
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411058416.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Traditional security systems are difficult to identify specific security needs, resulting in unreasonable resource allocation, unreasonable equipment deployment, and lack of efficient coordination mechanisms, which affects monitoring and early warning efficiency.

Method used

Provide a smart security early warning device, including security requirements reading module, deployment decision module, chip configuration module, edge AI fusion module and early warning management module. By reading the security requirements of industrial parks, equipment deployment and data are deployed in combination with intelligent decision model. Flow architecture design, configuration of AI vision chips to realize edge AI convergence and collaborative early warning management of distributed devices.

Benefits of technology

Ensure the reasonable allocation of security resources, avoid monitoring blind spots, realize efficient edge-side monitoring and real-time early warning management, ensure low latency and high reliability of data transmission, and improve the timeliness of security incident detection and response.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007363A_ABST
    Figure CN119007363A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent security early warning device and method for an industrial park, and relates to the technical field of security early warning, and the device comprises a security demand reading module which is used for reading the security demand of a target industrial park; the deployment decision module is used for determining a data stream architecture; the chip configuration module is used for configuring an AI visual chip; the edge AI fusion module is used for deploying an intelligent security and protection system by carrying out edge AI fusion on the distributed equipment; the early warning management module is used for carrying out edge security early warning management and central control cooperative early warning management; and the transmission management module is used for carrying out edge side monitoring and autonomous decision response, carrying out security interaction video stream transmission management based on a network time protocol, and carrying out security tracking through interaction neighborhood equipment. According to the invention, the technical problems of low accuracy and efficiency of security early warning caused by unreasonable distribution of security resources and lack of an efficient cooperation mechanism between security devices in the prior art are solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of security warning technology, and in particular to a smart security warning device and method for an industrial park. Background Art

[0002] In response to the demand for smart security and early warning for large facilities such as industrial parks, traditional security systems have many technical problems. On the one hand, the security needs of different application scenarios and facilities are different, but traditional security systems find it difficult to accurately identify specific security needs, resulting in irrational resource allocation and ineffective protection. In addition, the deployment of security equipment in traditional security systems often relies on manual experience and lacks scientific basis, resulting in irrational equipment distribution and monitoring blind spots. On the other hand, the edge devices in traditional security systems lack an efficient collaboration mechanism, and cannot achieve real-time data sharing and collaborative work between multiple devices, affecting the overall efficiency of monitoring and early warning. Summary of the invention

[0003] This application provides a smart security warning device and method for industrial parks, aiming to solve the technical problems that the deployment of security equipment in the prior art is often based on experience, cannot be targeted at specific application scenarios and security needs, resulting in unreasonable allocation of security resources, and lacks an efficient coordination mechanism between security equipment, resulting in low accuracy and efficiency of security warnings.

[0004] The first aspect disclosed in the present application provides a smart security warning device for an industrial park, and the method includes: a security demand reading module, the security demand reading module is used to read the security demand of the target industrial park; a deployment decision module, the deployment decision module is used to make deployment decisions based on the security demand and in combination with the intelligent decision model, and determine the data flow architecture based on the security distribution; a chip configuration module, the chip configuration module is used to configure the AI ​​vision chip based on the data flow architecture; an edge AI fusion module, the edge AI fusion module is used to deploy the smart security system based on the AI ​​vision chip by performing edge AI fusion on distributed devices; an early warning management module, the early warning management module is used to perform edge security early warning management based on the distributed devices based on the smart security system, and collaborate with the central control for early warning management, wherein the distributed devices are interactive; a transmission management module, the transmission management module is used to perform security interactive video stream transmission management based on the network time protocol as the distributed devices perform edge-side monitoring and autonomous decision-making responses, wherein security tracking is performed through interactive neighborhood devices.

[0005] The second aspect disclosed in the present application provides a smart security warning method for industrial parks, and the device is used for the above-mentioned smart security warning method for industrial parks, and the device includes: reading the security needs of the target industrial park; guided by the security needs, making deployment decisions in combination with the intelligent decision-making model, and determining the data flow architecture based on security distribution; based on the data flow architecture, configuring AI vision chips; based on the AI ​​vision chip, deploying smart security systems by edge AI fusion of distributed devices; based on the smart security system, edge security warning management based on the distributed devices is performed, and collaborative warning management with the central control is performed, wherein the distributed devices are interactive; as the distributed devices perform edge-side monitoring and autonomous decision-making responses, security interactive video stream transmission management is performed based on the network time protocol, wherein security tracking is performed through interactive neighborhood devices.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] Through the security demand reading module, the specific security needs of the target industrial park can be accurately read and analyzed to ensure the rational allocation and effective protection of security resources; the deployment decision module combines the intelligent decision-making model to make deployment decisions, and determines the data flow architecture based on security distribution to ensure reasonable equipment distribution and no monitoring blind spots; the chip configuration module optimizes the configuration of AI vision chips based on the data flow architecture, simplifies equipment configuration and management, and meets complex and changeable security needs; the edge AI fusion module deploys smart security systems by integrating edge AI with distributed devices to achieve efficient edge-side monitoring and real-time early warning management; the early warning management module is based on the edge security early warning management of distributed devices, and collaborates with the central control for early warning management to ensure the timeliness of security event detection and response, and quickly handle emergencies; the transmission management module performs video stream transmission management for security interaction based on the network time protocol to ensure low latency and high reliability of data transmission, and realizes security tracking through interactive neighborhood devices to comprehensively protect the safety of the industrial park.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the structure of a smart security warning device for an industrial park provided in an embodiment of the present application;

[0010] Figure 2 A flow chart of a smart security warning method for an industrial park provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: security requirement reading module 10, deployment decision module 20, chip configuration module 30, edge AI fusion module 40, early warning management module 50, transmission management module 60. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a smart security warning method for industrial parks, thereby solving the technical problems that the deployment of security equipment in the prior art is often based on experience and cannot be targeted at specific application scenarios and security needs, resulting in unreasonable allocation of security resources, and lack of efficient coordination mechanism between security equipment, resulting in low accuracy and efficiency of security warnings.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] like Figure 1 As shown, the embodiment of the present application provides a smart security warning device for an industrial park, the device comprising:

[0015] A security requirement reading module 10, wherein the security requirement reading module 10 is used to read the security requirements of the target industrial park;

[0016] Read the security needs of the target industrial park. For example, by establishing communications with management personnel and security managers, interactively obtain opinions and suggestions on the security protection of the target industrial park, including security areas, key facilities, personnel flow, etc.; or, by analyzing the past security incident records of the industrial park, conduct risk assessments, and determine the high-risk areas and key protection targets of the park. Sort out the acquired security needs to form a comprehensive list of needs, prioritize the needs according to their urgency and importance, and clarify which needs need to be met first, laying the foundation for subsequent steps.

[0017] A deployment decision module 20, which is used to make deployment decisions based on the security requirements and in combination with an intelligent decision model, and to determine a data flow architecture based on security distribution;

[0018] Select a suitable intelligent decision-making model, such as a decision tree model based on machine learning, and use security requirements and historical data as input data into the intelligent decision-making model for model training. Specifically, extract features related to security requirements, such as regional risk level, historical event frequency, equipment coverage, etc. Use historical data to train the model, optimize model parameters, and improve the model's prediction accuracy. Through training on historical data, the model can predict the best deployment plan under different security requirements.

[0019] Based on the input data and prediction results, the model generates specific deployment decisions, including the deployment location, quantity, and configuration parameters of each security device, such as recommending the deployment of more monitoring equipment in high-risk areas. Analyze the security distribution in the industrial park, use distributed computing technology to distribute security data between edge devices and central control systems, reduce latency, and improve response speed. Based on the data flow relationship between each security device, determine the data flow architecture based on security distribution to ensure efficient transmission of data flows, including data interaction between edge devices and central control systems.

[0020] A chip configuration module 30, the chip configuration module 30 is used to configure the AI ​​vision chip based on the data flow architecture;

[0021] Furthermore, the chip configuration module 30 configures the AI ​​vision chip based on the data flow architecture, including the following steps:

[0022] The data flow architecture is traversed to determine effective security targets and invalid backgrounds, wherein effective security targets of various security devices are differentiated; the effective security targets are clustered to determine a plurality of clustering clusters; the plurality of clustering clusters are traversed to perform intra-class standardization, and the AI ​​vision chip is configured based on standardized security warning targets.

[0023] Collect real-time video streams from distributed devices and perform preliminary data processing such as denoising and filtering to ensure data quality. Use image processing and computer vision technology to extract features from video streams, such as the shape, color, and speed of moving objects, and establish a background model. Through time series analysis, distinguish between stable background and dynamic targets, identify valid security targets such as vehicles and people, and ignore invalid background such as trees swaying in the wind.

[0024] Extract features from the identified effective security targets, construct feature vectors, standardize the feature vectors, and eliminate the dimensional differences between different features. Use clustering algorithms, such as K-means, DBSCAN, etc., to perform cluster analysis on the feature vectors and determine multiple clusters, such as vehicles of different models and colors.

[0025] Since different distributed devices have different monitoring and early warning requirements and cannot be standardized uniformly, it is necessary to standardize each cluster within the cluster to ensure that each type of target is effectively processed in its application scenario. Specifically, multiple clusters are obtained from the cluster analysis results, each cluster contains a specific type of security target, and the security targets in each cluster are traversed to collect the feature data of each target. The feature data of the security targets in the cluster are standardized so that they can be compared at the same scale. For example, each feature value is standardized by Z-score, that is, the mean is subtracted and then divided by the standard deviation.

[0026] Based on the standardized processing results within the class, set standardized security warning targets, such as detection thresholds and alarm strategies for different types of targets. Optimize the algorithm model in the AI ​​vision chip for standardized targets of different clusters, such as adjusting the number of layers and parameters of the convolutional neural network. According to processing requirements, adjust the hardware configuration of the AI ​​vision chip, such as improving GPU computing power, optimizing memory management, etc., to improve the accuracy and speed of target detection.

[0027] An edge AI fusion module 40, wherein the edge AI fusion module 40 is used to deploy a smart security system by performing edge AI fusion on distributed devices based on the AI ​​vision chip;

[0028] Identify distributed devices, such as smart cameras, sensors, edge computing nodes, etc., which can be seamlessly integrated with AI vision chips, formulate equipment deployment strategies based on security needs and data flow architecture, and ensure that important areas and high-risk areas have sufficient equipment coverage.

[0029] AI vision chips are embedded in distributed devices, so that each device has independent data processing and intelligent analysis capabilities. Optimized image processing and behavior analysis algorithms are deployed on each distributed device, so that the device can perform real-time data processing and threat detection. All distributed devices are connected through the network to form a collaborative smart security system, and real-time data and analysis results are shared between devices to improve the overall system's response speed and decision-making accuracy.

[0030] An early warning management module 50, which is used to perform edge security early warning management based on the distributed devices based on the smart security system, and coordinate early warning management with the central control, wherein the distributed devices are interactive;

[0031] Distributed devices use embedded AI vision chips to monitor and analyze video streams in real time and detect potential security threats. When an abnormal event is detected, the device immediately generates warning information, including detailed information such as event type, location, and time. Distributed devices exchange information through the network, collaborate in analysis and response, and improve the accuracy and timeliness of warnings.

[0032] Each distributed device transmits the warning information to the central control system, which summarizes and further analyzes the information, makes comprehensive decisions, and guides the distributed devices to take corresponding response measures, such as dispatching more equipment for monitoring and alarming. The central control system feeds back the decision results to the distributed devices and adjusts their monitoring and warning strategies in real time to ensure the efficient operation of the overall system.

[0033] Furthermore, the warning management module 50, the central control collaborative warning management, includes the following operation steps:

[0034] Obtain central processing tasks, pre-process video streams to determine video fragments, and transmit them to the control center; based on the control center, perform security decision positioning on the video fragments to determine security events; if the security event meets the decision accuracy, generate warning information based on the security event.

[0035] The control center assigns processing tasks to distributed devices according to real-time monitoring requirements. Distributed devices receive task instructions and perform video stream preprocessing. Specifically, the continuous video stream is decomposed into several video segments, or key frames are extracted from the video stream to reduce the amount of data and retain important information. Video fragments are obtained based on the processing results for subsequent processing. The video fragments are transmitted to the control center through a network connection.

[0036] The control center receives video fragments transmitted from distributed devices and analyzes them using AI algorithms. For example, it uses deep learning models to analyze videos in real time to detect security events. It combines the geographic information in the video to accurately locate the location of the security event and determines the specific time of the security event through the timestamp of the video frame.

[0037] Use multiple verification algorithms, such as cross-validation and confidence calculation, to verify the accuracy of security events, such as verifying whether the confidence of a suspicious person entering a restricted area exceeds a preset threshold, so as to determine whether the warning conditions are met. If they are met, integrate detailed information about the security event, including event type, location, time, etc., generate warning information, and send the warning information to relevant departments to ensure timely response.

[0038] Furthermore, the warning management module 50 further includes the following operation steps:

[0039] If the security event does not meet the decision-making accuracy, feedback monitoring information is generated, and the feedback monitoring information identifies a collaborative device group. Based on the feedback monitoring information, the collaborative device group is subjected to the same-target collaborative monitoring and decision-making; wherein, the determination of the collaborative device group includes: reading the distribution of neighboring devices of the feedback device; using the full-view monitoring as the screening requirement, and using the monitoring of overlapping domains and idle states as the constraints, and screening and determining the collaborative device group based on the distribution of neighboring devices.

[0040] If the security event does not meet the decision-making accuracy, detect whether there are analysis errors caused by factors such as angle, light, target occlusion, etc., generate feedback monitoring information including current analysis results, accuracy assessment, reasons for abnormal detection, etc., and determine nearby equipment groups with collaborative capabilities based on the monitoring range of the current device and the distribution of neighboring devices.

[0041] The generated feedback monitoring information is sent to the identified collaborative device group. The collaborative devices receive the feedback information, parse it and conduct collaborative monitoring of the same target, that is, real-time monitoring of the same target from multiple angles and perspectives is performed. The monitoring data and analysis results are shared among the collaborative devices to improve the accuracy of the overall analysis.

[0042] Based on the monitoring data of multiple devices, comprehensive analysis and decision-making are carried out. For example, multi-view fusion algorithms, such as multi-view stereo vision and feature fusion methods, are used to integrate the monitoring data of each device, improve the accuracy of target recognition, and determine the true situation of security events. If a security event is confirmed, high-precision warning information is generated and released.

[0043] Identify the devices that currently have analysis errors or low precision, namely the backhaul devices, confirm the specific location of the backhaul devices through the device location information, query the distribution of other monitoring devices around the backhaul devices, namely the neighboring devices, from the device management system, and integrate the location information, monitoring range, current status and other data of the neighboring devices.

[0044] Use full monitoring perspective as the screening requirement to ensure that the selected collaborative device group can cover the perspectives that the backhaul device cannot monitor. Determine the constraints, including monitoring overlap and idle status. Specifically, try to select devices with low monitoring overlap to ensure that collaborative devices can provide new monitoring perspectives; give priority to devices that are currently idle to avoid affecting other monitoring tasks.

[0045] Analyze the monitoring range, overlapping area, current status and other information of neighboring devices, select neighboring devices that meet the requirements according to the set screening requirements and constraints, and determine the collaborative device group. Specifically, calculate the monitoring perspective and coverage of each device, determine the perspective overlap, select devices that meet the full perspective coverage, query the current status of the device, give priority to idle or low-load devices, and form a collaborative device group based on the selected devices, thereby achieving efficient collaborative monitoring and decision-making with the same goal.

[0046] Furthermore, the early warning management module 50, the pre-processing of the video stream to determine the video fragments, comprises the following operation steps:

[0047] By performing edge recognition, key frame images are extracted from the video stream, wherein a preset edge movement interval is used as a key frame determination standard; image calibration processing and image fuzzy segmentation processing are performed on the key frame images to determine the video fragments, wherein the image processing requirement is multiple image defects, and the image segmentation standard is effective security targets and invalid backgrounds.

[0048] Obtain real-time video streams from distributed monitoring devices, use algorithms in AI vision chips, such as the Canny edge detection algorithm, to detect edges in the video stream, identify edges in the image, set a preset edge movement interval, and identify edge areas that may contain target events. By detecting edge movement, select the image frame with the largest edge change as the key frame image for extraction.

[0049] Perform image calibration on key frame images, including geometric calibration to correct image tilt and distortion, and color calibration to adjust image color and brightness to ensure image consistency and visibility. Use fuzzy segmentation technology to segment key frame images, extract areas that may contain target events, and perform denoising and enhancement on multiple defects in the image, such as noise and fuzzy areas, to improve image quality. According to the image segmentation standard, distinguish effective security targets from invalid backgrounds, and identify image segments containing effective security targets as video fragments for subsequent analysis.

[0050] The transmission management module 60 is used to manage the transmission of video streams for security interaction based on the network time protocol as the distributed device performs edge-side monitoring and autonomous decision-making responses, wherein security tracking is performed through interactive neighborhood devices.

[0051] Distributed devices monitor the environment in real time at the edge, analyze video streams through built-in AI vision chips, detect potential security threats, and make autonomous decisions on detected threats based on preset algorithms, generating early warning information or taking corresponding measures, such as alarms and tracking.

[0052] All distributed devices are synchronized through the network time protocol, which ensures the timing consistency of data transmission between different devices and avoids data confusion caused by clock differences. Based on the network time protocol, each device transmits the real-time monitoring video stream to the central control system or other devices. When a device detects abnormal behavior, the neighboring device immediately receives the collaborative information and starts synchronous monitoring and tracking. Through the collaboration of multiple devices, the behavior trajectory of the threat target is analyzed and predicted to ensure comprehensive monitoring and accurate tracking.

[0053] Furthermore, the transmission management module 60 performs video stream transmission management of security interaction based on the network time protocol, including the following operation steps:

[0054] Traverse the distributed devices and timestamp the monitoring video stream; read the collaborative monitoring task, and based on the network time protocol, perform synchronous transmission management of the associated data stream based on the collaborative monitoring task, wherein the data transmission includes the edge side and the control center, and between the edge sides; wherein the network time protocol can perform transmission collision coordination management.

[0055] Traverse all distributed devices, obtain monitoring video streams from each distributed device, synchronize the time of all distributed devices, and insert accurate timestamps into each frame of the video stream to ensure that each frame of the video has a corresponding timestamp.

[0056] The control center assigns collaborative monitoring tasks to each distributed device based on real-time monitoring needs. The distributed devices receive and parse the collaborative monitoring tasks and perform data transmission, including data transmission between the edge side and the control center and data interaction between edge devices. Specifically, on the one hand, edge devices collect real-time video streams and analyze data, and transmit video streams and data to the control center through the network to ensure the real-time and integrity of the data. On the other hand, after a device detects a security warning target, it notifies nearby successor devices to monitor according to the target's trajectory. Edge devices share monitoring data, collaborate on analysis, and make decisions.

[0057] Furthermore, the transmission management module 60, the security tracking through the interactive neighborhood device, includes the following operation steps:

[0058] Based on the first distribution device, field of view monitoring and video stream analysis are performed to determine a first warning target, where the first distribution device is any device among the distributed devices; first warning information based on the first warning target is generated and an alarm is given, a behavior trajectory of the first warning target is predicted, and first prediction information is determined; based on the first prediction information, a second distribution device is located, where the second distribution device is a trajectory deflection device in the vicinity of the first distribution device; a first tracking signal is generated and transmitted to the second distribution device, so as to track and monitor the first warning target.

[0059] Select any distributed device as the first distributed device, which monitors its field of view in real time and captures the video stream. Use AI vision chips and target detection algorithms to analyze the video stream and detect the first warning target.

[0060] According to the detected first warning target, the warning information containing the target location, time, type and other information is generated, and the warning information is sent to the monitoring center through audio, video or SMS for immediate warning. The motion trajectory of the first warning target is analyzed by using the multi-frame data in the video stream, and the future motion trajectory of the target is predicted, and the first prediction information containing the predicted target location, time, etc. is generated for subsequent decision-making.

[0061] According to the first prediction information, the predicted trajectory of the warning target is obtained, its possible movement direction and target predicted position are determined, the distribution of neighboring devices of the first distribution device is queried, the device covering the target predicted position is found, and the trajectory deviation device is selected from the neighboring devices, that is, the second distribution device that can best cover the target predicted position.

[0062] A first tracking signal including a first warning target is generated, and the second distribution device receives the first tracking signal, locks and tracks the first warning target according to the first tracking signal, captures the movement of the target in real time, and ensures continuous monitoring.

[0063] In summary, the smart security warning device for industrial parks provided in the embodiments of the present application has the following technical effects:

[0064] Through the security demand reading module 10, the specific security needs of the target industrial park are accurately read and analyzed to ensure the reasonable allocation and effective protection of security resources; the deployment decision module 20 makes deployment decisions in combination with the intelligent decision-making model, determines the data flow architecture based on security distribution, and ensures that the equipment is reasonably distributed and there are no monitoring blind spots; the chip configuration module 30 optimizes the configuration of AI vision chips based on the data flow architecture, simplifies equipment configuration and management, and meets complex and changeable security needs; the edge AI fusion module 40 deploys a smart security system by integrating edge AI with distributed devices to achieve efficient edge-side monitoring and real-time early warning management; the early warning management module 50 is based on the edge security early warning management of distributed devices, and cooperates with the central control for early warning management to ensure the timeliness of security event detection and response, and quickly handle emergencies; the transmission management module 60 performs video stream transmission management for security interaction based on the network time protocol to ensure low latency and high reliability of data transmission, and realizes security tracking through interactive neighborhood devices to comprehensively protect the safety of the industrial park.

[0065] Based on the same inventive concept as the smart security warning device for industrial parks in the aforementioned embodiment, Figure 2 As shown, the embodiment of the present application provides a smart security early warning method for an industrial park, the method comprising:

[0066] Read the security needs of the target industrial park; based on the security needs, make deployment decisions in combination with the intelligent decision-making model, and determine the data flow architecture based on security distribution; based on the data flow architecture, configure the AI ​​vision chip; based on the AI ​​vision chip, deploy the smart security system by integrating edge AI with distributed devices; based on the smart security system, perform edge security early warning management based on the distributed devices, and collaborate with the central control for early warning management, wherein the distributed devices are interactive; as the distributed devices perform edge-side monitoring and autonomous decision-making responses, perform security interactive video stream transmission management based on the network time protocol, wherein security tracking is performed through interactive neighborhood devices.

[0067] Furthermore, based on the data flow architecture, the AI ​​vision chip is configured, including:

[0068] The data flow architecture is traversed to determine effective security targets and invalid backgrounds, wherein effective security targets of various security devices are differentiated; the effective security targets are clustered to determine a plurality of clustering clusters; the plurality of clustering clusters are traversed to perform intra-class standardization, and the AI ​​vision chip is configured based on standardized security warning targets.

[0069] Furthermore, based on the network time protocol, video stream transmission management for security interaction includes:

[0070] Traverse the distributed devices and timestamp the monitoring video stream; read the collaborative monitoring task, and based on the network time protocol, perform synchronous transmission management of the associated data stream based on the collaborative monitoring task, wherein the data transmission includes the edge side and the control center, and between the edge sides; wherein the network time protocol can perform transmission collision coordination management.

[0071] Furthermore, the central control collaborative early warning management includes:

[0072] Obtain central processing tasks, pre-process video streams to determine video fragments, and transmit them to the control center; based on the control center, perform security decision positioning on the video fragments to determine security events; if the security event meets the decision accuracy, generate warning information based on the security event.

[0073] Furthermore, the method comprises:

[0074] If the security event does not meet the decision-making accuracy, feedback monitoring information is generated, and the feedback monitoring information identifies a collaborative device group; based on the feedback monitoring information, the collaborative device group is subjected to the same-target collaborative monitoring and decision-making; wherein, the determination of the collaborative device group includes: reading the distribution of neighboring devices of the feedback device; using the full-view monitoring as the screening requirement, and using the monitoring of overlapping domains and idle states as the constraints, and screening and determining the collaborative device group based on the distribution of neighboring devices.

[0075] Furthermore, the preprocessing of the video stream to determine the video fragments includes:

[0076] By performing edge recognition, key frame images are extracted from the video stream, wherein a preset edge movement interval is used as a key frame determination standard; image calibration processing and image fuzzy segmentation processing are performed on the key frame images to determine the video fragments, wherein the image processing requirement is multiple image defects, and the image segmentation standard is effective security targets and invalid backgrounds.

[0077] Furthermore, the security tracking through interactive neighborhood devices includes:

[0078] Based on the first distribution device, field of view monitoring and video stream analysis are performed to determine a first warning target, where the first distribution device is any device among the distributed devices; first warning information based on the first warning target is generated and an alarm is given, a behavior trajectory of the first warning target is predicted, and first prediction information is determined; based on the first prediction information, a second distribution device is located, where the second distribution device is a trajectory deflection device in the vicinity of the first distribution device; a first tracking signal is generated and transmitted to the second distribution device, so as to track and monitor the first warning target.

[0079] Through the above-mentioned detailed description of a smart security warning device for industrial parks in this specification, technical personnel in this field can clearly understand a smart security warning method for industrial parks in this embodiment. Since it corresponds to the device disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0080] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A smart security warning device for industrial parks, characterized in that: The device comprises: A security demand reading module, which is used to read the security demand of the target industrial park; A deployment decision module, which is used to make deployment decisions based on the security requirements and in combination with an intelligent decision model to determine a data flow architecture based on security distribution; A chip configuration module, the chip configuration module is used to configure the AI ​​vision chip based on the data flow architecture; An edge AI fusion module, which is used to deploy a smart security system by performing edge AI fusion on distributed devices based on the AI ​​vision chip; An early warning management module, which is used to perform edge security early warning management based on the distributed devices based on the smart security system and coordinate early warning management with the central control, wherein the distributed devices are interactive; A transmission management module is used to manage the transmission of video streams for security interaction based on the network time protocol as the distributed device performs edge-side monitoring and autonomous decision-making responses, wherein security tracking is performed through interactive neighborhood devices.

2. The device according to claim 1, characterized in that The chip configuration module includes the following steps: Traversing the data flow architecture to determine effective security targets and invalid backgrounds, wherein effective security targets of various security devices are different; Clustering the effective security targets to determine a plurality of clusters; The multiple clusters are traversed, and intra-class standardization is performed, and the AI ​​vision chip is configured based on standardized security warning targets.

3. The device according to claim 1, characterized in that The transmission management module includes the following steps: Traversing the distributed devices and timestamping the surveillance video stream; Reading the collaborative monitoring task, and executing the synchronous transmission management of the associated data stream based on the collaborative monitoring task based on the network time protocol, wherein the data transmission includes the edge side and the control center, and between the edge sides; Wherein, the network time protocol can perform transmission collision coordination management.

4. The device according to claim 3, characterized in that The early warning management module includes the following steps: Obtaining central processing tasks, pre-processing video streams to determine video fragments, and transmitting them to the control center; Based on the control center, security decision-making and positioning are performed on the video fragments to determine security events; If the security event meets the decision-making accuracy, early warning information based on the security event is generated.

5. The device according to claim 4, characterized in that The early warning management module also includes the following operation steps: If the security event does not meet the decision accuracy, feedback monitoring information is generated, and the feedback monitoring information is identified with a collaborative device group; Based on the feedback monitoring information, the collaborative device group is monitored and decided on the same target collaboratively; The determination of the collaborative device group includes: Read the distribution of neighboring devices of the backhaul device; The collaborative device group is determined based on the distribution of the neighboring devices by taking the monitoring of the full viewing angle as the screening requirement and taking the monitoring of the overlapping domain and the idle state as the constraint conditions.

6. The device according to claim 4, characterized in that The early warning management module also includes the following operation steps: By performing edge recognition, extracting key frame images from the video stream, wherein a preset edge movement interval is used as a key frame determination standard; The key frame images are subjected to image calibration processing and image fuzzy segmentation processing to determine the video fragments, wherein the image processing requirement is multi-image defects, and the image segmentation standard is effective security targets and invalid backgrounds.

7. The device according to claim 1, characterized in that The transmission management module includes the following steps: Based on a first distributed device, visual field monitoring and video stream analysis are performed to determine a first warning target, wherein the first distributed device is any device among the distributed devices; Generate first warning information based on the first warning target and issue a warning, predict the behavior trajectory of the first warning target, and determine first prediction information; Based on the first prediction information, locate a second distribution device, where the second distribution device is a trajectory deflection device in the neighborhood of the first distribution device; A first tracking signal is generated and transmitted to the second distribution device to track and monitor the first warning target.

8. A smart security early warning method for industrial parks, characterized in that: The method is implemented by a smart security warning device for an industrial park according to any one of claims 1 to 7, and the method comprises: Read the security requirements of the target industrial park; Guided by the security requirements, deployment decisions are made in combination with intelligent decision-making models to determine the data flow architecture based on security distribution; Based on the data flow architecture, configure the AI ​​vision chip; Based on the AI ​​vision chip, a smart security system is deployed by integrating edge AI into distributed devices; Based on the smart security system, edge security early warning management based on the distributed devices is performed, and early warning management is coordinated with the central control, wherein the distributed devices are interactive; As the distributed devices perform edge-side monitoring and autonomous decision-making responses, security interactive video stream transmission management is performed based on the network time protocol, wherein security tracking is performed through interactive neighborhood devices.

Citation Information

Cited By

  • Digital intelligent residential district security decision-making system and method based on end-side-cloud collaboration

    CN119672928A

  • Intelligent early warning method and system for safety risk of campus

    CN120525360A

  • A smart early warning method and system for campus safety risks

    CN120525360B

  • Intelligent inspection management and control system integrating multi-mode security protection and unmanned aerial vehicle cooperation

    CN121262341A