A campus security all-weather monitoring system and monitoring method based on digital twin model

Abstract

The present invention discloses a park security all-weather monitoring system and monitoring method based on a digital twin model, which relates to the field of security monitoring technology. The system acquires and stores video streams and environmental data in real time through a data acquisition module, providing a basis for subsequent analysis. The central processing module preprocesses and extracts features of the data to generate a video image feature set and an environmental feature set. The digital twin monitoring module visualizes environmental features through a three-dimensional model to achieve an intuitive display of the dynamic changes in the chemical workshop. The video quality analysis module uses the video quality index Q(t) and the monitoring and control index C(t) to comprehensively calculate the comprehensive monitoring performance index Ezh to ensure the clarity and stability of video monitoring. The comprehensive evaluation and early warning module combines the abnormal detection value A(t) with the comprehensive monitoring performance index Ezh, calculates the comprehensive safety risk index O(t) and generates early warning information to improve the safety management level of the park.

Description

Technical Field

[0001] The present invention relates to the field of security monitoring technology, and in particular to a campus security all-weather monitoring system and a monitoring method based on a digital twin model. Background Art

[0002] With the continuous advancement of modern technology, security monitoring systems, as an important tool to ensure public safety and private asset security, are becoming an important part of smart city construction. The application areas of this system cover many aspects from urban public security to industrial parks. Especially in special environments such as chemical parks that require high security protection, the role of security monitoring systems is particularly important. Due to the complexity and potential dangers of the production process, chemical parks need to achieve all-weather real-time monitoring and accurate environmental perception to ensure production safety and prevent accidents.

[0003] At present, traditional chemical park monitoring systems mainly rely on cameras and sensors for real-time monitoring. However, these systems usually lack integration and are not easy to effectively combine video data with environmental parameters, resulting in insufficient recognition and early warning capabilities for complex environmental conditions in production workshops in chemical parks. In addition, traditional systems are often not easy to adjust monitoring parameters in a timely manner when faced with environmental changes such as changes in lighting, smoke and chemical gas leaks, thus affecting monitoring quality. This limitation leads to insufficient response speed and accuracy when security incidents occur, increasing security risks. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a campus security all-weather monitoring system and monitoring method based on a digital twin model, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: including a data acquisition module, a central processing module, a digital twin monitoring module, a video quality analysis module and a comprehensive evaluation and early warning module;

[0006] The data acquisition module is used to integrate the monitoring cameras in the production workshop of the chemical park to obtain the video stream in real time, and at the same time, collect environmental data in real time based on the sensor group installed in the workshop and store it;

[0007] The central processing module is used to extract video streams and environmental data in real time, perform preprocessing to obtain a video image data set and an environmental data set, and then perform feature extraction to obtain a video image feature set and an environmental feature set;

[0008] The digital twin monitoring module is used to construct a digital twin model of a chemical workshop, integrate the acquired video image feature set and environmental feature set, and perform visual dynamic environment simulation;

[0009] The video quality analysis module is used to perform dimensionless processing on the acquired video image feature set and environmental feature set, and then perform correlation calculation to obtain the video quality index Q(t) and the monitoring and control index C(t), and perform comprehensive summary calculation to obtain the comprehensive monitoring performance index Ezh, and preset the monitoring performance threshold X for preliminary comparative evaluation to analyze the quality of the monitoring video;

[0010] The comprehensive evaluation and early warning module is used to perform calculation and analysis based on the video image feature set to obtain the abnormal detection value A(t), and then perform comprehensive calculation with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t), and then perform a secondary comparative evaluation between the preset safety threshold A and the obtained comprehensive safety risk index O(t), and generate early warning information based on the evaluation results.

[0011] Preferably, the data acquisition module includes a video stream acquisition unit, an environmental data acquisition unit and a data storage unit;

[0012] The video stream acquisition unit uses a surveillance camera deployed in the chemical park workshop to collect the video stream of the chemical park workshop in real time at 30 frames per second;

[0013] The environmental data collection unit collects environmental data in real time based on a sensor group deployed in a workshop in a chemical park;

[0014] The sensor group includes a compound gas sensor, a light sensor, a temperature sensor, an electrochemical corrosion sensor, a spectroscopic reflectometer, an electromagnetic field detector and a sound level meter;

[0015] The environmental data include light intensity Gz, spectral reflectance Gp, ambient temperature Wd, corrosive gas concentration Fs, volatile organic compound concentration Cg, electromagnetic interference index Im and ambient noise Zs;

[0016] The data storage unit is used to set up a wireless 5G network with a surveillance camera and a sensor group, establish a network connection, receive the collected video stream and environmental data in real time, and build a database to store the collected video stream and environmental data.

[0017] Preferably, the central processing module includes a data processing unit and a feature extraction unit;

[0018] The data processing unit is used to build an AIP application program interface and integrate it with the database, extract the collected video stream and environmental data in real time, and use video analysis software to extract video frame images from each frame of the video, and perform image stitching and calibration on each frame of the multi-source video stream to obtain a video image data group to ensure the consistency and seamlessness of the video stream; then use a mean filter on the environmental data to remove random errors in the environmental data, then adjust the environmental data to zero mean and unit variance, unify the data scales of different sensors, and then integrate and summarize the processed environmental data to generate an environmental data group;

[0019] The feature extraction unit includes a video feature extraction unit and an environment feature extraction unit;

[0020] The video feature extraction unit is used to extract color features, texture features, shape features, operation features and high-level features in the image based on the preprocessed video image data group using computer vision technology, and summarize them to generate video image content features, then perform feature fusion and clustering on the generated video image content features to obtain video content feature values ​​Vi, and then use computer vision technology to extract illumination features in the image, the illumination features include image brightness Ld and environmental exposure Bg, and then perform time series processing on the extracted video content features and illumination features to extract video content feature values ​​Vi(t) at time t, image brightness Ld(t) at time t, and environmental exposure Bg(t) at time t;

[0021] The color features are used to describe the overall distribution of colors by counting the distribution of RGB color components in the image, converting the image from the RGB space to other color spaces HSV and Lab to extract color features that are more in line with human eye perception;

[0022] The texture feature analyzes the spatial relationship between pixel gray levels through the gray level co-occurrence matrix to extract texture information, and then describes the local texture features through the local binary pattern, which is particularly suitable for tasks such as face recognition, and then uses wavelet transform frequency domain analysis to extract multi-scale texture features of the image;

[0023] The shape feature detects and analyzes the contour shape of the object in the image by using edge detection algorithms such as Sobel and Canny to identify the edge contour of the object and extract shape features such as area and perimeter.

[0024] The motion feature describes the direction and speed of the object's motion by calculating the motion vector field in the video picture feature set, and then extracts the foreground area of ​​the moving object through background modeling and differentiation;

[0025] The high-level features detect specific objects such as people and vehicles in the image by using deep learning methods such as convolutional neural networks (CNNs), extract the category and location of the target, and identify and classify specific behavior patterns by analyzing the action features in the video frame sequence;

[0026] By using color features and texture features to monitor environmental changes, it can identify environmental changes in early fire smoke. By analyzing motion features and shape features, it can identify abnormal behavior of unauthorized personnel entering sensitive areas. By extracting and analyzing high-level features, it can realize intelligent monitoring functions such as face recognition and license plate recognition.

[0027] The environmental feature extraction unit is used to align the timestamps of the environmental data collected by sensor groups at different times through time series processing technology to generate an environmental feature set, wherein the environmental feature set includes the light intensity Gz(t) at time t, the spectral reflectance Gp(t) at time t, the ambient temperature Wd(t) at time t, the corrosive gas concentration Fs(t) at time t, the volatile organic compound concentration Cg(t) at time t, the electromagnetic interference index Im(t) at time t, and the ambient noise Zs(t) at time t.

[0028] Preferably, the digital twin monitoring module is used to fuse the acquired video image feature set and environmental feature set to obtain environmental perception. First, it is ensured that the video frame and sensor data have accurate timestamps to achieve data synchronization, and the environmental feature set is matched with the monitoring area of ​​the video image feature set to establish a mapping of the physical space. The information from different data sources is merged using an algorithm to form a more comprehensive description of the environment. Then, 3D modeling software is used to establish a three-dimensional model of the chemical park workshop, including the equipment layout, spatial layout and channels in the workshop, and real-time video data and sensor data are mapped to the three-dimensional model to achieve dynamic environmental display. Physical simulation engines Unity and UnrealEngine are used to simulate the dynamic situation in the chemical park workshop, including personnel flow and environmental changes. Through the API application program interface, the real-time data stream and the three-dimensional model are connected to integrate and build a digital twin model, and an early warning prompt function is set.

[0029] Preferably, the video quality analysis module includes a video quality analysis unit, a camera setting unit and a comprehensive monitoring analysis unit;

[0030] Therefore, the video quality analysis unit performs correlation calculations through the extracted video image feature set and environmental feature set to obtain the video quality index Q(t) to analyze the quality of video surveillance;

[0031] The video quality index Q(t) is calculated by the following algorithm formula:

[0032]

[0033] In the formula, β 1 represents the influence coefficient of the volatile organic compound concentration Cg(t) at time t, β 2 The influence coefficient of the electromagnetic interference index Im(t) at time t is expressed. This formula integrates the impact of various environmental factors on video quality. By considering parameters such as brightness, exposure, gas concentration, electromagnetic interference and environmental noise, the formula can evaluate the quality changes of real-time video.

[0034] Preferably, the camera setting unit performs correlation calculation to obtain the monitoring and control index C(t) through the extracted video image feature set and environmental feature set;

[0035] The monitoring and control index C(t) is calculated by the following algorithm formula:

[0036] C(t)=α·[Gz(t)·(1+γ·Gp(t))]+δ·[Wd(t)·(1+η·Fs(t))];

[0037] Wherein, α·[Gz(t)·(1+γ·Gp(t))] is a component used to adjust the exposure value of the camera setting to cope with changes in different ambient lighting conditions and scene reflection characteristics, γ represents the influence coefficient of the spectral reflectance Gp(t) at time t, α represents the illumination change adjustment coefficient, δ·[Wd(t)·(1+η·Fs(t))] dynamically adjusts the camera setting by combining the factors of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t, δ represents the adjustment coefficient of the sum of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t, and η represents the adjustment coefficient of the corrosive gas concentration Fs(t) at time t.

[0038] Preferably, the comprehensive monitoring and analysis unit includes a comprehensive monitoring performance calculation unit and a performance evaluation unit;

[0039] The comprehensive monitoring performance calculation unit is used to correlate the obtained video quality index Q(t) and the monitoring control index C(t) to obtain the comprehensive monitoring performance index Ezh, quantify the overall performance of the monitoring system, analyze the matching degree between the video quality and the camera control, and judge whether the working state of the system is normal within a certain period of time;

[0040] The comprehensive monitoring performance index Ezh is calculated by the following algorithm formula:

[0041]

[0042] In the formula, T represents the monitoring time period;

[0043] The performance evaluation unit performs a preliminary comparative evaluation with the obtained comprehensive monitoring performance index Ezh by using the performance index of historical monitoring normal events to analyze the performance index of the monitoring system;

[0044] The specific evaluation plan is as follows:

[0045] When the comprehensive monitoring performance index Ezh> monitoring performance threshold X, it means that the current monitoring system performance is abnormal. The reasons include video quality problems: such as blurred images, excessive noise, underexposure or overexposure, etc., improper camera settings: such as failure to adapt to environmental conditions such as changes in lighting and temperature in time, resulting in inappropriate setting parameters, and environmental factors: external environments such as strong light, smoke and chemical gases have a negative impact on video quality. At this time, the optimization strategy is executed for analysis and early warning;

[0046] When the comprehensive monitoring performance index Ezh ≤ the monitoring performance threshold X, it indicates that the current monitoring system is operating normally and the existing settings are maintained to continue monitoring.

[0047] Preferably, the comprehensive assessment and early warning module includes an abnormal behavior detection unit, a security analysis unit and an assessment and early warning unit;

[0048] The abnormal behavior detection unit is used to extract the video content feature value Vi(t) at time t in the video image feature set in real time, calculate and obtain the abnormal detection value A(t), and analyze the abnormal behavior of the monitored video area;

[0049] The abnormal detection value A(t) is calculated and obtained by the following algorithm formula:

[0050]

[0051] Where Wi represents the weight coefficient of the i-th frame video content, N represents the total number of frames, and λ represents the influence coefficient of the environmental parameters, which is used to adjust the influence of the environmental parameters on the anomaly detection value;

[0052] represents the average value of video content at time t;

[0053] This item represents the square of the deviation between the feature value of the video content of the i-th frame and the average feature value of the video content. The square operation amplifies the detection of abnormal behaviors or states that deviate from the normal state. If the feature value of a frame deviates significantly from the average value, it means that the frame may contain abnormal information, such as sudden movement, the appearance or disappearance of an object, etc.

[0054] (1+λ·(Cg(t)+Im(t))) represents the environmental parameter influencing factor, which reflects the impact of environmental parameters on video content. Environmental changes such as increased gas concentration or enhanced electromagnetic interference may cause fluctuations in video quality, so the sensitivity of anomaly detection needs to be adjusted. This factor plays an amplifying or reducing role in the formula. If the environmental parameters change greatly, such as high gas concentration or strong electromagnetic interference, the video quality may deteriorate. In this case, the anomaly detection value needs a higher sensitivity;

[0055] The safety analysis unit is used to associate the acquired abnormal detection value A(t) with the comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t);

[0056] The comprehensive safety risk index O(t) is calculated by the following algorithm formula:

[0057]

[0058] Where θ represents the anomaly detection influence coefficient, M represents the total number of different areas or different types of abnormal behaviors monitored within time t, j represents the index variable, which is used to iteratively calculate the detection value of each abnormal behavior in each area, and κ represents the optimization coefficient, which is used to adjust the baseline of the optimized adjustment value of the final video monitoring;

[0059] It represents the comprehensive anomaly detection value, which is the sum of all abnormal behaviors detected at time t.

[0060] Preferably, the evaluation and early warning unit performs a preset safety threshold A based on the safety requirements and operation strategies of the park, and then performs a secondary comparative evaluation with the obtained comprehensive safety risk index O(t), analyzes the current security monitoring situation in the workshop of the chemical park, and generates early warning information;

[0061] The specific evaluation plan is as follows:

[0062] When the comprehensive security risk index O(t) ≥ the security threshold A, it means that the video quality of the current monitoring system is abnormal, which may be due to changes in environmental factors such as strong light or smoke. At the same time, multiple abnormal behaviors are detected. At this time, an early warning message is sent through the digital twin model to indicate the existence of security anomalies;

[0063] When the comprehensive security risk index O(t) is less than the security threshold A, it means that the video quality is within the normal range and no minor abnormalities are detected. In this case, there is no need to generate warning information.

[0064] A method for all-weather monitoring of campus security based on a digital twin model includes the following steps:

[0065] S1. First, the surveillance cameras in the production workshop of the chemical park are integrated to obtain the video stream in real time. At the same time, the environmental data is collected and stored in real time based on the sensor group installed in the workshop;

[0066] S2, then integrate with the database by building an API application program interface, extract the video stream and environmental data stored in the database in real time, perform preprocessing to obtain a video image data set and an environmental data set, and then perform feature extraction to obtain a video image feature set and an environmental feature set;

[0067] S3. By building a digital twin model of a chemical workshop, the acquired video image feature set and environmental feature set are integrated to perform a visual dynamic environment simulation;

[0068] S4, after dimensionless processing based on the acquired video image feature set and environmental feature set, the video quality index Q(t) and the monitoring and control index C(t) are obtained by correlation calculation, and a comprehensive summary calculation is performed to obtain the comprehensive monitoring performance index Ezh, and the monitoring performance threshold X is preset for preliminary comparative evaluation to analyze the quality of the monitoring video;

[0069] S5. Finally, based on the video image feature set, calculation and analysis are performed to obtain the anomaly detection value A(t), which is then combined with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t). The preset safety threshold A is then compared and evaluated with the obtained comprehensive safety risk index O(t) for a second time, and warning information is generated based on the evaluation results.

[0070] The present invention provides a park security all-weather monitoring system and monitoring method based on a digital twin model. It has the following beneficial effects:

[0071] (1) The system can acquire high-definition video streams and environmental data in the production workshop of the chemical park in real time through the integrated data acquisition module. The video stream acquisition unit works together with the environmental data acquisition unit to record the video data and environmental data in the production workshop of the chemical park in real time. The central processing module extracts and preprocesses these data, and the digital twin monitoring module uses these data to build a three-dimensional digital twin model of the park and perform dynamic environmental simulation. This comprehensive environmental monitoring capability enables the system to promptly identify and respond to emergencies such as gas leaks or equipment failures, and ensures rapid and accurate response measures through real-time data updates and dynamic display of the three-dimensional model.

[0072] (2) The system calculates the video quality index Q(t) and the monitoring and control index C(t) through the video quality analysis module to comprehensively evaluate the quality of video surveillance. The video quality index Q(t) comprehensively considers the impact of various environmental factors on video quality, thereby accurately reflecting the clarity and stability of real-time video. The monitoring and control index C(t) adjusts the camera settings according to factors such as lighting, ambient temperature, and corrosive gas concentration to cope with different environmental conditions. This precise video quality analysis and dynamic control capability ensures that high-quality video surveillance data can be obtained under various environmental conditions, improving the reliability and monitoring effect of the system.

[0073] (3) The system uses the comprehensive evaluation and early warning module to analyze the video image feature values ​​and environmental data in real time through the abnormal behavior detection unit and the security risk assessment unit, and calculate the abnormal detection value A(t) and the comprehensive security risk index O(t). The abnormal detection value A(t) identifies possible abnormal behaviors, such as sudden movement or the appearance of objects, by analyzing the deviation of the video content characteristics. The comprehensive security risk index O(t) combines the abnormal detection value with the comprehensive monitoring performance index Ezh to conduct a comprehensive risk assessment. When the comprehensive security risk index O(t) exceeds the preset security threshold A, the system can generate an early warning message to indicate potential security anomalies. This intelligent anomaly detection and security early warning mechanism greatly improves the efficiency of park security management, timely discovers and handles safety hazards, and ensures the safe operation of the park. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The above and other objects, features and advantages of the present application will become more apparent through a more detailed description of exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0075] Figure 1 This is a schematic diagram of the process of a park security all-weather monitoring system based on a digital twin model of the present invention;

[0076] Figure 2 This is a schematic diagram of the steps of a method for all-weather monitoring of campus security based on a digital twin model in the present invention. DETAILED DESCRIPTION

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] Example 1

[0079] See also Figure 1 The present invention provides a park security all-weather monitoring system based on a digital twin model. To achieve the above purpose, the present invention is implemented through the following technical solutions: including a data acquisition module, a central processing module, a digital twin monitoring module, a video quality analysis module and a comprehensive evaluation and early warning module;

[0080] The data acquisition module is used to integrate the surveillance cameras in the production workshop of the chemical park to obtain the video stream in real time. At the same time, it collects and stores the environmental data in real time based on the sensor group installed in the workshop.

[0081] The central processing module is used to extract video streams and environmental data in real time, perform preprocessing to obtain video image data sets and environmental data sets, and then perform feature extraction to obtain video image feature sets and environmental feature sets;

[0082] The digital twin monitoring module is used to build a digital twin model of the chemical workshop, integrating the acquired video image feature set and environmental feature set to perform visual dynamic environment simulation;

[0083] The video quality analysis module is used to perform dimensionless processing based on the acquired video image feature set and environmental feature set, and then perform correlation calculation to obtain the video quality index Q(t) and the monitoring and control index C(t), and perform comprehensive summary calculation to obtain the comprehensive monitoring performance index Ezh, and preset the monitoring performance threshold X for preliminary comparative evaluation to analyze the quality of the monitoring video;

[0084] The comprehensive evaluation and early warning module is used to perform calculations and analysis based on the video image feature set to obtain the anomaly detection value A(t), and then perform a comprehensive calculation with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t), and then perform a secondary comparative evaluation between the preset safety threshold A and the obtained comprehensive safety risk index O(t), and generate early warning information based on the evaluation results.

[0085] In this embodiment, the system integrates monitoring cameras and environmental sensors through the data acquisition module, acquires and stores video streams and environmental data in real time, and provides rich raw data for subsequent analysis. The central processing module is responsible for preprocessing and feature extraction of these data, converting video image data and environmental data into video image feature sets and environmental feature sets with analytical value, laying a solid foundation for subsequent digital twin model construction and quality analysis. The digital twin monitoring module visualizes video image features and environmental features by constructing a three-dimensional digital twin model of the chemical workshop, and provides dynamic environmental simulation. The digital twin monitoring module makes it possible to map real-time monitoring data with virtual models, thereby realizing the intuitive display and simulation of various dynamic changes in the chemical workshop. The video quality analysis module further improves the intelligence level of the system. By calculating the video quality index Q(t) and the monitoring and control index C(t), a comprehensive summary calculation is performed to obtain the comprehensive monitoring performance index Ezh, comprehensively evaluate the quality of the monitoring video and the setting of the camera, and ensure the clarity and stability of video monitoring under different environmental conditions. The comprehensive evaluation and early warning module combines the abnormal detection value A(t) and the comprehensive monitoring performance index Ezh, calculates the comprehensive safety risk index O(t), performs a secondary evaluation and generates early warning information. This module effectively improves the system's safety early warning capabilities, can identify potential safety risks in real time and issue early warnings, and significantly improves the safety management level of the park.

[0086] Example 2

[0087] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a video stream acquisition unit, an ,environmental data acquisition unit and a data storage unit;

[0088] The video stream acquisition unit uses surveillance cameras deployed in the chemical park workshop to collect video streams of the chemical park workshop in real time at 30 frames per second;

[0089] The environmental data collection unit collects environmental data in real time based on the sensor group deployed in the workshop of the chemical park;

[0090] The sensor group includes compound gas sensors, light sensors, temperature sensors, electrochemical corrosion sensors, spectroscopic reflectometers, electromagnetic field detectors, and sound level meters;

[0091] Environmental data include light intensity Gz, spectral reflectance Gp, ambient temperature Wd, corrosive gas concentration Fs, volatile organic compound concentration Cg, electromagnetic interference index Im and ambient noise Zs;

[0092] The data storage unit is used to set up the wireless 5G network with the surveillance camera and sensor group, establish a network connection, receive the collected video stream and environmental data in real time, and build a database to store the collected video stream and environmental data.

[0093] In this embodiment, the system constructs an efficient and comprehensive data acquisition and storage system through the data acquisition module through its video stream acquisition unit, environmental data acquisition unit and data storage unit. The video stream acquisition unit uses 30 frames per second real-time video acquisition technology to ensure the continuity and clarity of high frame rate video data, and provides real-time visual information for the monitoring system. The environmental data acquisition unit comprehensively collects environmental data through a variety of sensors. These data provide detailed environmental background for subsequent analysis, helping the system to better understand and respond to different environmental changes. The data storage unit uses a wireless 5G network to achieve real-time connection with surveillance cameras and sensor groups, ensuring instant transmission and reliable storage of data. This efficient storage solution not only improves the data transmission speed, but also ensures the integrity and security of the data.

[0094] Example 3

[0095] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the central processing module includes a data processing unit and a feature extraction unit;

[0096] The data processing unit is used to build an AIP application program interface for integration with the database, extract the collected video stream and environmental data in real time, and use video analysis software to extract video frame images from each frame of the video, and perform image stitching and calibration on each frame of the multi-source video stream to obtain a video image data group to ensure the consistency and seamlessness of the video stream; then use a mean filter on the environmental data to remove random errors in the environmental data, and then adjust the environmental data to zero mean and unit variance, unify the data scales of different sensors, and then integrate and summarize the processed environmental data to generate an environmental data group;

[0097] The feature extraction unit includes a video feature extraction unit and an environment feature extraction unit;

[0098] The video feature extraction unit is used to extract color features, texture features, shape features, operation features and high-level features in the image based on the preprocessed video image data group by applying computer vision technology, and summarize them to generate video image content features, and then perform feature fusion and clustering on the generated video image content features to obtain video content feature values ​​Vi, and then use computer vision technology to extract illumination features in the image, the illumination features include image brightness Ld and environmental exposure Bg, and then perform time series processing on the extracted video content features and illumination features to extract the video content feature value Vi(t) at time t, the image brightness Ld(t) at time t and the environmental exposure Bg(t) at time t;

[0099] Color features, which are used to describe the overall distribution of colors by counting the distribution of RGB color components in the image, convert the image from RGB space to other color spaces HSV and Lab, and extract color features that are more in line with human eye perception;

[0100] Texture features use the gray-level co-occurrence matrix to analyze the spatial relationship between pixel gray levels, extract texture information, and then describe local texture features through local binary patterns, which is particularly suitable for tasks such as face recognition. Wavelet transform frequency domain analysis is then used to extract multi-scale texture features of the image;

[0101] Shape features: By using edge detection algorithms such as Sobel and Canny to identify the edge contours of objects, detect and analyze the contour shapes of objects in images, and extract shape features such as area and perimeter.

[0102] The motion feature describes the direction and speed of the object's motion by calculating the motion vector field in the video image feature set, and then extracts the foreground area of ​​the moving object through background modeling and differentiation;

[0103] High-level features detect specific objects such as people and vehicles in images by using deep learning methods such as convolutional neural networks (CNNs), extract the category and location of the target, and identify and classify specific behavior patterns by analyzing the action features in video frame sequences;

[0104] By using color features and texture features to monitor environmental changes, it can identify environmental changes in early fire smoke. By analyzing motion features and shape features, it can identify abnormal behavior of unauthorized personnel entering sensitive areas. By extracting and analyzing high-level features, it can realize intelligent monitoring functions such as face recognition and license plate recognition.

[0105] The environmental feature extraction unit is used to align the timestamps of the environmental data collected by sensor groups at different times through time series processing technology to generate an environmental feature set, which includes the light intensity Gz(t) at time t, the spectral reflectance Gp(t) at time t, the ambient temperature Wd(t) at time t, the corrosive gas concentration Fs(t) at time t, the volatile organic compound concentration Cg(t) at time t, the electromagnetic interference index Im(t) at time t and the ambient noise Zs(t) at time t.

[0106] Example 4 This example is an explanation of Example 1. Figure 1 Specifically: the digital twin monitoring module is used to fuse the acquired video image feature set and environmental feature set to obtain environmental perception. First, ensure that the video frame and sensor data have accurate timestamps to achieve data synchronization, correspond the environmental feature set to the monitoring area of ​​the video image feature set, establish a physical space mapping, and use algorithms to merge information from different data sources to form a more comprehensive environmental description; then use 3D modeling software to establish a three-dimensional model of the chemical park workshop, including the equipment layout, space layout and channels in the workshop, and then map real-time video data and sensor data to the three-dimensional model to achieve dynamic environmental display, use physical simulation engines Unity and UnrealEngine to simulate the dynamic situation in the chemical park workshop, including personnel flow and environmental changes, and connect the real-time data stream and the three-dimensional model through the API application program interface to integrate and build a digital twin model, and set an early warning prompt function.

[0107] In this embodiment, the system is seamlessly integrated with the database through the AIP application program interface constructed by the data processing unit, and extracts and preprocesses the video stream and environmental data in real time. Each frame of the video stream is spliced ​​and calibrated to ensure the continuity and consistency of the data. The environmental data is filtered by a mean value to remove random errors and standardized to a uniform scale, providing accurate environmental background information for subsequent analysis. This refined data processing not only improves the reliability of the data, but also provides a solid foundation for feature extraction. The feature extraction unit further deepens the data analysis and uses advanced computer vision technology to extract rich features from the video frames, including color features, texture features, shape features, motion features and high-level features. At the same time, the time series processing of environmental data also generates a detailed set of environmental features. Color features optimize the description of human eye perception through color space conversion, texture features provide detailed texture information through grayscale co-occurrence matrix and wavelet transform, and shape features and motion features help identify object contours and motion patterns. High-level feature extraction realizes intelligent monitoring functions such as face and license plate recognition. These features not only enhance the monitoring capability of environmental changes, but also improve the detection accuracy of abnormal behavior and security events. The digital twin monitoring module builds a three-dimensional digital model of the chemical park workshop by integrating the video image feature set with the environmental feature set. This model provides a real-time dynamic environmental display and simulates the actual situation inside the workshop using 3D modeling and physical simulation technology.

[0108] Example 4

[0109] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the video quality analysis module includes a video quality ,analysis unit, a camera setting unit and a comprehensive monitoring ,analysis unit;

[0110] Therefore, the video quality analysis unit performs correlation calculations through the extracted video image feature set and environmental feature set to obtain the video quality index Q(t) to analyze the quality of video surveillance;

[0111] The video quality index Q(t) is calculated using the following algorithm formula:

[0112]

[0113] In the formula, β 1 represents the influence coefficient of the volatile organic compound concentration Cg(t) at time t, β 2 The influence coefficient of the electromagnetic interference index Im(t) at time t is expressed. This formula integrates the impact of various environmental factors on video quality. By considering parameters such as brightness, exposure, gas concentration, electromagnetic interference and environmental noise, the formula can evaluate the quality changes of real-time video.

[0114] The camera setting unit performs correlation calculation to obtain the monitoring and control index C(t) through the extracted video image feature set and environmental feature set;

[0115] The monitoring and control index C(t) is calculated by the following algorithm formula:

[0116] C(t)=α·[Gz(t)·(1+γ·Gp(t))]+δ·[Wd(t)·(1+η·Fs(t))];

[0117] Wherein, α·[Gz(t)·(1+γ·Gp(t))] is a component used to adjust the exposure value of the camera setting to cope with changes in different ambient lighting conditions and scene reflection characteristics, γ represents the influence coefficient of the spectral reflectance Gp(t) at time t, α represents the illumination change adjustment coefficient, δ·[Wd(t)·(1+η·Fs(t))] dynamically adjusts the camera setting by combining the factors of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t, δ represents the adjustment coefficient of the sum of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t, and η represents the adjustment coefficient of the corrosive gas concentration Fs(t) at time t.

[0118] The comprehensive monitoring and analysis unit includes a comprehensive monitoring performance calculation unit and a performance evaluation unit;

[0119] The comprehensive monitoring performance calculation unit is used to correlate the obtained video quality index Q(t) and the monitoring control index C(t) to obtain the comprehensive monitoring performance index Ezh, quantify the overall performance of the monitoring system, analyze the matching degree between the video quality and the camera control, and judge whether the working state of the system is normal within a certain period of time;

[0120] The comprehensive monitoring performance index Ezh is calculated by the following algorithm formula:

[0121]

[0122] In the formula, T represents the monitoring time period;

[0123] The performance evaluation unit uses the performance indicators of historical monitoring normal events to preset the monitoring performance threshold X, and then conducts a preliminary comparative evaluation with the obtained comprehensive monitoring performance indicator Ezh to analyze the performance indicators of the monitoring system;

[0124] The specific evaluation plan is as follows:

[0125] When the comprehensive monitoring performance index Ezh> monitoring performance threshold X, it means that the current monitoring system performance is abnormal. The reasons include video quality problems: such as blurred images, excessive noise, underexposure or overexposure, etc., improper camera settings: such as failure to adapt to environmental conditions such as changes in lighting and temperature in time, resulting in inappropriate setting parameters, and environmental factors: external environments such as strong light, smoke and chemical gases have a negative impact on video quality. At this time, the optimization strategy is executed for analysis and early warning;

[0126] When the comprehensive monitoring performance index Ezh ≤ the monitoring performance threshold X, it indicates that the current monitoring system is operating normally and the existing settings are maintained to continue monitoring.

[0127] In this embodiment, the system uses the video image feature set and the environmental feature set through the video quality analysis unit to calculate the video quality index Q(t) through a complex algorithm formula, and comprehensively considers the impact of environmental factors on video quality. This process ensures that the quality of real-time video monitoring can be accurately evaluated, providing a reliable basis for subsequent regulation. The camera setting unit calculates the monitoring and control index C(t) through the extracted feature data, and dynamically adjusts the camera settings to adapt to different environmental lighting and scene changes. This unit optimizes the exposure value of the camera based on factors such as ambient temperature, corrosive gas concentration and spectral reflectivity, thereby improving the accuracy and clarity of video capture. This real-time adjustment capability greatly enhances the flexibility of the system when facing different environmental conditions. The comprehensive monitoring and analysis unit combines the video quality index Q(t) and the monitoring and control index C(t) to calculate the comprehensive monitoring performance index Ezh, and quantitatively evaluates the overall performance of the monitoring system. This module can not only perform a preliminary evaluation of the system based on historical data and performance threshold X, but also identify the causes of abnormal system performance, such as video quality problems, improper camera settings or environmental factors.

[0128] Example 5

[0129] This embodiment is explained in Example 4. Please refer to Figure 1 ,Specifically: the comprehensive evaluation and early warning module includes an abnormal ,behavior detection unit, a security analysis unit and an evaluation and early ,warning unit;

[0130] The abnormal behavior detection unit is used to extract the video content feature value Vi(t) at time t in the video image feature set in real time, calculate and obtain the abnormal detection value A(t), and analyze the abnormal behavior of the monitored video area;

[0131] The anomaly detection value A(t) is calculated by the following algorithm formula;

[0132]

[0133] Where Wi represents the weight coefficient of the i-th frame video content, N represents the total number of frames, and λ represents the influence coefficient of the environmental parameters, which is used to adjust the influence of the environmental parameters on the anomaly detection value;

[0134] represents the average value of video content at time t;

[0135] This item represents the square of the deviation between the feature value of the video content of the i-th frame and the average feature value of the video content. The square operation amplifies the detection of abnormal behaviors or states that deviate from the normal state. If the feature value of a frame deviates significantly from the average value, it means that the frame may contain abnormal information, such as sudden movement, the appearance or disappearance of an object, etc.

[0136] (1+λ·(Cg(t)+Im(t))) represents the environmental parameter influencing factor, which reflects the impact of environmental parameters on video content. Environmental changes such as increased gas concentration or enhanced electromagnetic interference may cause fluctuations in video quality, so the sensitivity of anomaly detection needs to be adjusted. This factor plays an amplifying or reducing role in the formula. If the environmental parameters change greatly, such as high gas concentration or strong electromagnetic interference, the video quality may deteriorate. In this case, the anomaly detection value needs a higher sensitivity;

[0137] The safety analysis unit is used to correlate the obtained abnormal detection value A(t) with the comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t);

[0138] The comprehensive safety risk index O(t) is calculated by the following algorithm formula:

[0139]

[0140] Where θ represents the anomaly detection influence coefficient, M represents the total number of different areas or different types of abnormal behaviors monitored within time t, j represents the index variable, which is used to iteratively calculate the detection value of each abnormal behavior in each area, and κ represents the optimization coefficient, which is used to adjust the baseline of the optimized adjustment value of the final video monitoring;

[0141] It represents the comprehensive anomaly detection value, which is the sum of all abnormal behaviors detected at time t.

[0142] The evaluation and early warning unit presets the safety threshold A based on the safety requirements and operation strategies of the park, and then conducts a secondary comparative evaluation with the obtained comprehensive safety risk index O(t), analyzes the current security monitoring situation in the workshop of the chemical park, and generates early warning information;

[0143] The specific evaluation plan is as follows:

[0144] When the comprehensive security risk index O(t) ≥ the security threshold A, it means that the video quality of the current monitoring system is abnormal, which may be due to changes in environmental factors such as strong light or smoke. At the same time, multiple abnormal behaviors are detected. At this time, an early warning message is sent through the digital twin model to indicate the existence of security anomalies;

[0145] When the comprehensive security risk index O(t) is less than the security threshold A, it means that the video quality is within the normal range and no minor abnormalities are detected. In this case, there is no need to generate warning information.

[0146] In this embodiment, the abnormal behavior detection unit extracts and calculates the video content feature value Vi(t) in real time, and uses a specific algorithm formula to evaluate the abnormal detection value A(t). This process can keenly identify abnormal behaviors in video frames, such as sudden movements or abnormal changes of objects, and timely discover potential safety hazards. In addition, the adjustment of the abnormal detection value by the influence coefficient of environmental parameters further improves the accuracy of detection, so that the system can maintain efficient detection capabilities even when the environment changes greatly. The security analysis unit combines the abnormal detection value A(t) with the comprehensive monitoring performance index Ezh to calculate the comprehensive security risk index O(t). This unit provides an in-depth assessment of the overall security risk of the current monitoring system by comprehensively analyzing the abnormal detection results and monitoring performance. This comprehensive analysis method can more comprehensively reflect the security risks and optimize the monitoring system, thereby improving the ability to handle multiple abnormal behaviors in complex environments. The evaluation and early warning unit sets the security threshold A according to the security requirements and operation strategies of the park, and compares it with the comprehensive security risk index O(t) to generate early warning information.

[0147] Example 6

[0148] See also Figure 1 and Figure 2 , a campus security all-weather monitoring method based on a digital twin model, comprising the following steps:

[0149] S1. First, the surveillance cameras in the production workshop of the chemical park are integrated to obtain the video stream in real time. At the same time, the environmental data is collected and stored in real time based on the sensor group installed in the workshop;

[0150] S2, then integrate with the database by building an API application program interface, extract the video stream and environmental data stored in the database in real time, perform preprocessing to obtain a video image data set and an environmental data set, and then perform feature extraction to obtain a video image feature set and an environmental feature set;

[0151] S3. By building a digital twin model of a chemical workshop, the acquired video image feature set and environmental feature set are integrated to perform a visual dynamic environment simulation;

[0152] S4, after dimensionless processing based on the acquired video image feature set and environmental feature set, the video quality index Q(t) and the monitoring and control index C(t) are obtained by correlation calculation, and a comprehensive summary calculation is performed to obtain the comprehensive monitoring performance index Ezh, and the monitoring performance threshold X is preset for preliminary comparative evaluation to analyze the quality of the monitoring video;

[0153] S5. Finally, based on the video image feature set, calculation and analysis are performed to obtain the anomaly detection value A(t), which is then combined with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t). The preset safety threshold A is then compared and evaluated with the obtained comprehensive safety risk index O(t) for a second time, and warning information is generated based on the evaluation results.

[0154] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A campus security all-weather monitoring system based on a digital twin model, characterized by: It includes data acquisition module, central processing module, digital twin monitoring module, video quality analysis module and comprehensive evaluation and early warning module; The data acquisition module is used to integrate the monitoring cameras in the production workshop of the chemical park to obtain the video stream in real time, and at the same time, collect environmental data in real time based on the sensor group installed in the workshop and store it; The central processing module is used to extract video streams and environmental data in real time, perform preprocessing to obtain a video image data set and an environmental data set, and then perform feature extraction to obtain a video image feature set and an environmental feature set; The digital twin monitoring module is used to construct a digital twin model of a chemical workshop, integrate the acquired video image feature set and environmental feature set, and perform visual dynamic environment simulation; The video quality analysis module is used to perform dimensionless processing based on the acquired video image feature set and environmental feature set, and then perform correlation calculation to obtain the video quality index Q(t) and the monitoring and control index C(t), and perform comprehensive summary calculation to obtain the comprehensive monitoring performance index Ezh, and preset the monitoring performance threshold X for preliminary comparative evaluation to analyze the quality of the monitoring video; wherein, the video quality index Q(t) is calculated and obtained by the following algorithm formula: ； Wherein, Ld(t) represents the image brightness at time t, Bg(t) represents the environmental exposure at time t, Cg(t) represents the volatile organic compound concentration at time t, Im(t) represents the electromagnetic interference index at time t, and Zs(t) represents the environmental noise at time t. represents the influence coefficient of the volatile organic compound concentration Cg(t) at time t, Represents the influence coefficient of the electromagnetic interference index Im(t) at time t; the monitoring and control index C(t) is calculated and obtained by the following algorithm formula: ； Wherein, Gz(t) represents the light intensity at time t, Gp(t) represents the spectral reflectance at time t, Wd(t) represents the ambient temperature at time t, and Fs(t) represents the corrosive gas concentration at time t. A component used to adjust the exposure value of the camera setting to cope with changes in ambient lighting conditions and scene reflectivity characteristics. Represents the influence coefficient of the spectral reflectance Gp(t) at time t, Indicates the illumination change adjustment coefficient, By combining the factors of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t, the camera settings are dynamically adjusted. The adjustment coefficient represents the sum of the ambient temperature Wd(t) at time t and the corrosive gas concentration Fs(t) at time t. Indicates the adjustment coefficient of the corrosive gas concentration Fs(t) at time t; The comprehensive evaluation and early warning module is used to perform calculation and analysis based on the video image feature set to obtain the abnormal detection value A(t), and then perform comprehensive calculation with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t), and then perform a secondary comparative evaluation between the preset safety threshold A and the obtained comprehensive safety risk index O(t), and generate early warning information based on the evaluation results.

2. According to the digital twin model-based all-weather monitoring system for campus security according to claim 1, it is characterized by: The data acquisition module includes a video stream acquisition unit, an environmental data acquisition unit and a data storage unit; The video stream acquisition unit uses a surveillance camera deployed in the chemical park workshop to collect the video stream of the chemical park workshop in real time at 30 frames per second; The environmental data collection unit collects environmental data in real time based on a sensor group deployed in a workshop in a chemical park; The sensor group includes a compound gas sensor, a light sensor, a temperature sensor, an electrochemical corrosion sensor, a spectroscopic reflectometer, an electromagnetic field detector and a sound level meter; The environmental data include light intensity Gz, spectral reflectance Gp, ambient temperature Wd, corrosive gas concentration Fs, volatile organic compound concentration Cg, electromagnetic interference index Im and ambient noise Zs; The data storage unit is used to set up a wireless 5G network with a surveillance camera and a sensor group, establish a network connection, receive the collected video stream and environmental data in real time, and build a database to store the collected video stream and environmental data.

3. According to claim 2, a digital twin model-based park security all-weather monitoring system is characterized by: The central processing module includes a data processing unit and a feature extraction unit; The data processing unit is used to construct an AIP application program interface for integration with a database, extract the collected video stream and environmental data in real time, and use video analysis software to extract video frame images from each frame of the video, and perform image stitching and calibration on each frame of the multi-source video stream to obtain a video image data group; then use a mean filter on the environmental data to remove random errors in the environmental data, then adjust the environmental data to an average value and unit variance, unify the data scales of different sensors, and then integrate and summarize the processed environmental data to generate an environmental data group; The feature extraction unit includes a video feature extraction unit and an environment feature extraction unit; The video feature extraction unit is used to extract color features, texture features, shape features, operation features and high-level features in the image based on the preprocessed video image data group using computer vision technology, and summarize them to generate video image content features, and then perform feature fusion and clustering on the generated video image content features to obtain video content feature values ​​Vi, and then use computer vision technology to extract illumination features in the image, the illumination features include image brightness Ld and environmental exposure Bg, and then perform time series processing on the extracted video content features and illumination features to extract video content feature values ​​Vi(t) at time t, image brightness Ld(t) at time t, and environmental exposure Bg(t) at time t; The environmental feature extraction unit is used to align the timestamps of the environmental data collected by sensor groups at different times through a time series processing technology to generate an environmental feature set, wherein the environmental feature set includes the light intensity Gz(t) at time t, the spectral reflectance Gp(t) at time t, the ambient temperature Wd(t) at time t, the corrosive gas concentration Fs(t) at time t, the volatile organic compound concentration Cg(t) at time t, the electromagnetic interference index Im(t) at time t, and the ambient noise Zs(t) at time t.

4. According to claim 3, a digital twin model-based park security all-weather monitoring system is characterized by: The digital twin monitoring module is used to fuse the acquired video image feature set and environmental feature set to obtain environmental perception; Then use 3D modeling software to build a three-dimensional model of the chemical park workshop, including the equipment layout, space layout and channels in the workshop, and then map the real-time video data and sensor data to the three-dimensional model to achieve dynamic environmental display. Use the physical simulation engine to simulate the dynamic situation in the chemical park workshop, including personnel flow and environmental changes. Through the API application program interface, connect the real-time data stream and the three-dimensional model to integrate and build a digital twin model, and set up an early warning prompt function.

5. According to the digital twin model-based all-weather monitoring system for campus security according to claim 1, it is characterized by: The comprehensive monitoring and analysis unit includes a comprehensive monitoring performance calculation unit and a performance evaluation unit; The comprehensive monitoring performance calculation unit is used to correlate the obtained video quality index Q(t) and the monitoring control index C(t) to obtain the comprehensive monitoring performance index Ezh, and quantify the overall performance of the monitoring system; The comprehensive monitoring performance index Ezh is calculated by the following algorithm formula: ； In the formula, T represents the monitoring time period; The performance evaluation unit performs a preliminary comparative evaluation with the obtained comprehensive monitoring performance index Ezh by using the performance index of historical monitoring normal events to analyze the performance index of the monitoring system; The specific evaluation plan is as follows: When the comprehensive monitoring performance index Ezh>monitoring performance threshold X, it means that the current monitoring system performance is abnormal. At this time, the optimization strategy is executed for analysis and early warning; When the comprehensive monitoring performance index Ezh ≤ the monitoring performance threshold X, it indicates that the current monitoring system is operating normally and the existing settings are maintained to continue monitoring.

6. The digital twin model-based all-weather campus security monitoring system according to claim 1 is characterized by: The comprehensive assessment and early warning module includes an abnormal behavior detection unit, a security analysis unit and an assessment and early warning unit; The abnormal behavior detection unit is used to extract the video content feature value Vi(t) at time t in the video image feature set in real time, calculate and obtain the abnormal detection value A(t), and analyze the abnormal behavior of the monitored video area; The abnormal detection value A(t) is calculated and obtained by the following algorithm formula; ； Where Wi represents the weight coefficient of the i-th frame video content, N represents the total number of frames, Indicates the influence coefficient of environmental parameters, which is used to adjust the influence of environmental parameters on anomaly detection values; The safety analysis unit is used to correlate the acquired abnormal detection value A(t) with the comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t); The comprehensive safety risk index O(t) is calculated by the following algorithm formula: ； In the formula, represents the anomaly detection influence coefficient, M represents the total number of abnormal behaviors of different regions or different types monitored within time t, and j represents the index variable, which is used to iteratively calculate the detection value of each abnormal behavior in each region. Represents the optimization coefficient, which is used to adjust the baseline of the optimization adjustment value of the final video monitoring.

7. The digital twin model-based all-weather monitoring system for campus security according to claim 6 is characterized by: The evaluation and early warning unit presets a safety threshold A based on the safety requirements and operation strategies of the park, and then performs a secondary comparative evaluation with the obtained comprehensive safety risk index O(t), analyzes the current security monitoring situation in the workshop of the chemical park, and generates early warning information; The specific evaluation plan is as follows: When the comprehensive security risk index O(t) ≥ the security threshold A, it means that the video quality of the current monitoring system is abnormal, and multiple abnormal behaviors are detected at the same time. At this time, an early warning message is sent through the digital twin model to indicate the existence of security anomalies; When the comprehensive security risk index O(t) is less than the security threshold A, it means that the video quality is within the normal range and no minor abnormalities are detected. In this case, there is no need to generate warning information.

8. A method for all-weather monitoring of park security based on a digital twin model, applied to an all-weather monitoring system for park security based on a digital twin model as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: S1. First, the surveillance cameras in the production workshop of the chemical park are integrated to obtain the video stream in real time. At the same time, the environmental data is collected and stored in real time based on the sensor group installed in the workshop; S2, then integrate with the database by building an API application program interface, extract the video stream and environmental data stored in the database in real time, perform preprocessing to obtain a video image data set and an environmental data set, and then perform feature extraction to obtain a video image feature set and an environmental feature set; S3. By building a digital twin model of a chemical workshop, the acquired video image feature set and environmental feature set are integrated to perform a visual dynamic environment simulation; S4, after dimensionless processing based on the acquired video image feature set and environmental feature set, the video quality index Q(t) and the monitoring and control index C(t) are obtained by correlation calculation, and a comprehensive summary calculation is performed to obtain the comprehensive monitoring performance index Ezh, and the monitoring performance threshold X is preset for preliminary comparative evaluation to analyze the quality of the monitoring video; S5. Finally, based on the video image feature set, calculation and analysis are performed to obtain the anomaly detection value A(t), which is then combined with the obtained comprehensive monitoring performance index Ezh to obtain the comprehensive safety risk index O(t). The preset safety threshold A is then compared and evaluated with the obtained comprehensive safety risk index O(t) for a second time, and early warning information is generated based on the evaluation results.

Images