An AIoT-based intelligent security and protection early warning method and system

Abstract

The application relates to an AIoT-based intelligent security and protection early warning method and system, and relates to the field of security and protection, which comprises the following steps: collecting light detection information and temperature parameters in a natural environment in real time; generating environment adjustment information based on the light detection information and the temperature parameters and outputting the environment adjustment information to a collection device to collect multi-modal data; generating behavior information based on the multi-modal data; when the behavior information contains abnormal behavior, generating early warning information based on the environment adjustment information and the behavior information; triggering a hierarchical early warning mechanism based on the early warning information, and simultaneously encrypting and transmitting the early warning information to a security and protection system terminal to perform early warning. The application has the effect of reducing the missed judgment or misjudgment of a security and protection system.

Description

Technical Field

[0001] This invention relates to the field of security, and in particular to an intelligent security early warning method and system based on AIoT. Background Technology

[0002] Security is a systematic project that uses a combination of technical means, equipment deployment, and management systems to prevent, deter, and respond to various security threats, such as intrusion, theft, sabotage, and disasters, to protect the safety of people, property, information, and the environment.

[0003] Currently, traditional security systems typically deploy cameras in a dispersed manner at the locations requiring monitoring, and then transmit the monitoring images to the management room, relying on manual monitoring of the images in the management room.

[0004] Regarding the aforementioned technologies, traditional security systems rely on manual monitoring, which cannot maintain focused attention for extended periods, leading to missed or incorrect assessments and severely impacting the reliability of the security system. Summary of the Invention

[0005] To reduce missed or false alarms in security systems, this invention provides an intelligent security early warning method and system based on AIoT.

[0006] In a first aspect, the present invention provides an intelligent security early warning method based on AIoT, which adopts the following technical solution:

[0007] An AIoT-based intelligent security early warning method includes:

[0008] S1: Real-time acquisition of light detection information and temperature parameters in the natural environment;

[0009] S2: Generate environmental adjustment information based on the illumination detection information and the temperature parameters, and output it to the acquisition device to acquire multimodal data;

[0010] S3: Generate behavioral information based on the multimodal data;

[0011] S4: When abnormal behavior is detected in the behavioral information, an early warning message is generated based on the environmental adjustment information and the behavioral information;

[0012] S5: Based on the warning information, a graded warning mechanism is triggered, and the warning information is encrypted and transmitted to the security system terminal for warning purposes.

[0013] By adopting the above technical solution, the system dynamically optimizes the acquisition device settings by collecting and analyzing multimodal data through real-time acquisition of light and temperature parameters, thereby improving the accuracy and real-time performance of abnormal behavior identification in complex environments. At the same time, the acquired data is encrypted during transmission to enhance the security of information transmission.

[0014] Optionally, generating environmental adjustment information based on the illumination detection information and the temperature parameters further includes:

[0015] S21: When the light intensity in the light detection information is lower than the preset light reference intensity, obtain the rate of change of light occlusion, the shape characteristics of the occluded area and the movement trajectory within a preset time period from the light detection information.

[0016] S22: Establish a dynamic occlusion model based on the rate of change, the shape features, and the movement trajectory;

[0017] S23: Determine whether the dynamic occlusion model conforms to the preset artificial dynamic occlusion model;

[0018] S24: If the conditions are met, then based on the movement trajectory and the shape features, control the preset UAV to perform airspace fill-in and collect modal fill-in data, and at the same time, based on the movement trajectory, control the preset acquisition device to perform reverse displacement compensation, and use the collected data as compensation data.

[0019] S25: Generate spatiotemporal correlation data based on the modal complementation data and the compensation data;

[0020] S26: Generate spatiotemporal adjustment information based on the spatiotemporal correlation data and the temperature parameters, and use the spatiotemporal adjustment information as the environmental adjustment information;

[0021] S27: If it does not meet the requirements, obstacle avoidance information is generated based on the dynamic occlusion model;

[0022] S28: Generate environmental fusion information based on the obstacle avoidance information and the temperature parameters, and use the environmental fusion information as the environmental adjustment information.

[0023] Optionally, generating environmental adjustment information based on the illumination detection information and the temperature parameters further includes:

[0024] S291: When the light intensity of the light detection information is higher than the preset light reference intensity, the area with a light intensity higher than the preset light reference intensity is defined as a highlight area, and at the same time, it is determined whether the temperature parameter is higher than the preset temperature reference parameter.

[0025] S292: If the temperature is higher than the preset temperature reference parameter, the region above the preset temperature reference parameter is defined as a high temperature region;

[0026] S293: Generate a high-light-heat zone based on the high-light zone and the high-temperature zone;

[0027] S294: Obtain the current preset position of the light shield as the current position of the baffle;

[0028] S295: Calculate the rotation angle of the light-shielding plate based on the current position of the baffle and the high-light-heat zone as the adjustment angle;

[0029] S296: Generate device adjustment information based on the adjustment angle;

[0030] S297: Control the preset light-shielding plate to perform based on the device adjustment information, and simultaneously obtain the light adjustment detection information and temperature adjustment parameters after the execution;

[0031] S298: Generate an environmental regulation factor based on the illumination regulation detection information and the temperature regulation parameters, and use the environmental regulation factor as the environmental adjustment information;

[0032] S299: If the temperature is not higher than the preset temperature reference parameter, then retrieve the sound data from the multimodal data;

[0033] S29A: Optimize image acquisition parameters based on the illumination detection information;

[0034] S29B: Generate audio-visual adjustment information based on the image acquisition parameters and the sound data, and use the audio-visual adjustment information as the environmental adjustment information.

[0035] Optionally, generating device adjustment information based on the adjustment angle includes:

[0036] S2961: Obtain the monitoring area;

[0037] S2962: When the monitoring area is inconsistent with the preset monitoring area, the unmonitored area is named the monitoring blind zone;

[0038] S2963: The coordinate range of the monitoring blind zone is obtained based on a preset spatial coordinate system and used as the blind zone coordinates;

[0039] S2964: Calculate the difference in the monitoring area based on the blind zone coordinates;

[0040] S2965: Based on the difference in the monitored area, obtain the rotation angle difference and elevation angle difference, and control the light intensity sensor to perform the operation;

[0041] S2966: Generate a focus parameter correction amount based on the rotation angle difference, the pitch angle difference, and the blind zone coordinates;

[0042] S2967: Generate the device adjustment information based on the focus parameter correction amount, the rotation angle difference, and the pitch angle difference.

[0043] Optionally, methods for determining the presence of abnormal behavior include:

[0044] S61: Extract behavioral features from the multimodal data;

[0045] S62: When the behavioral features do not conform to the preset behavioral benchmark model, extract items that do not conform to the behavioral benchmark model from the behavioral features and mark them as abnormal items;

[0046] S63: Extract the number of the abnormal items and use it as the number of abnormal items;

[0047] S64: Generate anomaly confidence based on the anomalous items and the number of anomalous items;

[0048] S65: Determine whether the anomaly confidence level exceeds the preset anomaly baseline confidence level;

[0049] S66: If the limit is exceeded, it is determined that the abnormal behavior exists;

[0050] S67: If the number of occurrences of the abnormal item is not exceeded, the number of occurrences of the abnormal item is retrieved from the behavioral characteristics as the number of abnormal occurrences;

[0051] S68: If the number of anomalies exceeds a preset number, a second verification is performed based on the multimodal data.

[0052] Optionally, secondary verification based on the multimodal data includes:

[0053] S681: Retrieve sound data from the multimodal data;

[0054] S682: Generate acoustic pattern feature data based on the sound data and the behavioral features;

[0055] S683: Match the acoustic texture feature data with the preset acoustic texture database and extract the one with the highest matching degree as the initial anomaly judgment value;

[0056] S684: Calculate the difference between the initial anomaly judgment value and the benchmark matching degree as the anomaly difference;

[0057] S685: When the abnormal difference exceeds the preset abnormal difference range, it is determined that the abnormal behavior exists.

[0058] Optionally, methods for determining the presence of abnormal behavior may also include:

[0059] S71: Obtain the current pedestrian density from the multimodal data;

[0060] S72: Determine whether the current crowd density exceeds the preset density value;

[0061] S73: If the density value is exceeded, the movement trajectory, movement speed and distribution characteristics of the crowd are collected in real time;

[0062] S74: Generate group dynamic data based on the movement trajectory, the movement speed, and the distribution characteristics;

[0063] S75: Establish a crowd dynamic model based on the crowd dynamic data and the current crowd density, and label individual behavioral characteristics;

[0064] S76: Generate voice data based on the individual behavioral characteristics and the voice data;

[0065] S77: If the vocal data does not match the preset vocal baseline data within a preset time period, then an abnormal individual is marked and it is determined that the abnormal behavior exists.

[0066] Optionally, establishing a crowd dynamics model based on the crowd dynamics data and the current crowd density includes:

[0067] S751: Generate an initial dynamic model based on the aforementioned group dynamic data;

[0068] S752: Determine whether the initial dynamic model conforms to the group behavior baseline model;

[0069] S753: If they match, the initial dynamic model is adjusted based on the current crowd density to form a group dynamic model;

[0070] S754: If they do not match, extract the abnormal region from the initial dynamic model;

[0071] S755: Calculate the area change rate of the abnormal region within a preset time period to generate the diffusion rate of the abnormal region;

[0072] S756: Calculate the group risk index based on the diffusion speed of the abnormal area and the current population density;

[0073] S757: Establish a multi-level population model based on the initial dynamic model, the abnormal region, and the risk index, and use the multi-level population model as the dynamic population model.

[0074] Optionally, after encrypting and transmitting the warning information to the security system terminal for warning purposes, the system further includes:

[0075] S51: Generate a corrected risk index based on the environmental adjustment information and the multimodal data;

[0076] S52: Generate the rate of increase based on the modified risk index;

[0077] S53: When the rising rate is greater than the preset rising reference rate, the difference between the rising rate and the preset rising reference rate is calculated and used as the rising rate difference.

[0078] S54: Generate backup monitoring retrieval information based on the rise rate difference, and call the preset backup monitoring device for linkage based on the backup monitoring retrieval information;

[0079] S55: Generate early warning feedback information based on the modified risk index and the rate of increase;

[0080] S56: Encrypt the backup monitoring information and the early warning feedback information and update them synchronously to the multi-level early warning strategy library of the security system terminal.

[0081] Secondly, the present invention provides an intelligent security early warning system based on AIoT, which adopts the following technical solution:

[0082] An AIoT-based intelligent security early warning system includes:

[0083] The acquisition module is used to acquire illumination detection information, temperature parameters, and multimodal data;

[0084] A memory for storing an AIoT-based intelligent security early warning method as described in any one of the first aspects;

[0085] The processor loads and executes programs from memory.

[0086] In summary, this application includes at least one of the following beneficial technical effects:

[0087] 1. By dynamically optimizing the settings of the acquisition equipment through real-time collection of light and temperature parameters and combining this with the analysis of multimodal data, the accuracy and real-time performance of abnormal behavior identification in complex environments are significantly improved. At the same time, the acquired data is encrypted during transmission, which not only ensures the security of information transmission but also achieves efficient resource scheduling and rapid response.

[0088] 2. By identifying monitoring blind spots and establishing their spatial coordinates, the required rotation and elevation angle differences are calculated based on these coordinates. This, combined with focus parameter corrections, enables automatic calibration of the monitoring equipment. This achieves automatic elimination of blind spots, enhances the system's adaptability in complex environments, and significantly improves the efficiency of anomaly detection and the reliability of early warnings.

[0089] 3. Based on the initial dynamic model generated from crowd dynamic data, and compared with the behavioral baseline model, the spread rate of abnormal areas is identified. Combined with real-time pedestrian flow dynamics, a crowd risk index is calculated, enabling a quantitative assessment of potential security threats. By constructing a multi-layered dynamic model, the abnormal distribution and risk evolution trends are presented. This allows for optimization of model accuracy based on pedestrian flow under normal conditions, and precise location of risk sources, prediction of spread paths, and assessment of impact range in abnormal scenarios. This improves the accuracy and response efficiency of security early warning systems. Attached Figure Description

[0090] Figure 1 This is a flowchart of the intelligent security early warning method based on AIoT according to an embodiment of this application. Detailed Implementation

[0091] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0092] An AIoT-based intelligent security early warning method can improve the reliability of security systems by automatically identifying abnormal behaviors in different environments, thereby reducing missed or false alarms.

[0093] Reference Figure 1 This application discloses an AIoT-based intelligent security early warning method, which includes:

[0094] S1: Real-time acquisition of light detection information and temperature parameters in the natural environment.

[0095] The natural environment refers to the environment in which people live in daily life. Acquisition equipment refers to hardware that supports multimodal data acquisition, including image detection devices, sound sensors, and other devices. Illumination detection information refers to image information obtained by detecting illumination on the acquisition equipment. This information is continuously acquired through a matrix-set illumination intensity sensor pre-installed on the acquisition equipment and includes illumination intensity, rate of change, shape characteristics of occluded areas, and movement trajectory. Temperature parameters refer to the ambient temperature value, obtained through a temperature sensor. The acquisition of illumination detection information and temperature parameters in the environment provides specific data support for subsequent analysis.

[0096] S2: Generate environmental adjustment information based on illumination detection information and temperature parameters and output it to the acquisition device to collect multimodal data.

[0097] Environmental adjustment information refers to instructions for adjusting the acquisition equipment based on environmental conditions, such as adjusting the angle of the image detection device on the acquisition equipment and adjusting the focus parameters. Multimodal data refers to environmental data that integrates multiple dimensions such as vision and sound. Specific generation methods are detailed in S21 to S28. By adjusting the acquisition equipment under different environments before acquiring multimodal data, the integrity and quality of data in complex environments are ensured.

[0098] S3: Generate behavioral information based on multimodal data.

[0099] Behavioral information refers to the dynamics of a target object, such as its movement trajectory, speed, and body movements. Different multimodal data correspond to different behavioral information. Behavioral information is obtained by inputting multimodal data into a pre-set behavioral information database. The behavioral information database pre-stores a lookup table of different multimodal data and their corresponding behavioral information. The database is formed by technicians summarizing and recording the behaviors reflected by different multimodal data in sequence.

[0100] S4: When abnormal behavior is detected, an early warning message is generated based on the environmental adjustment information and the behavioral information.

[0101] Abnormal behavior refers to behavioral patterns that significantly deviate from normal behavioral characteristics, such as illegal intrusion, robbery, theft, and gang fights. Warning information refers to alarm signals that include information such as the type and specific location of the event. Methods for determining the presence of abnormal behavior are described in S61 to S68, S681 to S685, and S71 to S77. When abnormal behavioral information is detected, environmental adjustment information is analyzed. Combining this with the specific abnormal characteristics of the behavioral information, such as the type, frequency, and severity of the abnormal behavior, a comprehensive evaluation is conducted using pre-set warning rules and algorithm models to determine the warning level and content. This information is then integrated into a clear, accurate, and targeted warning message text, which is promptly sent to relevant responsible persons or management departments through pre-set communication channels such as SMS, email, and system push notifications, enabling them to quickly take appropriate countermeasures to ensure safety and stability. Accurate identification of abnormal events and their correlation with the environment improves intervention efficiency and enhances the accuracy of alarms. Warning rules and algorithm models are existing technologies and will not be elaborated upon here.

[0102] S5: Triggers a tiered early warning mechanism based on early warning information, and simultaneously encrypts and transmits the early warning information to the security system terminal for early warning.

[0103] A tiered early warning mechanism refers to a response level classification based on the severity of an event, such as Level 1, Level 2, and Level 3 alarms, with different alarm methods customizable. Encrypted transmission refers to end-to-end encryption of data. In this embodiment, the national cryptographic algorithm SM4 is used to encrypt the video stream, with a key dynamic update cycle of ≤5 minutes. The security system terminal refers to the central platform for receiving and processing alarms. Different early warning information triggers different tiered early warning mechanisms. The tiered early warning mechanism determines the triggered alarm level by querying a preset alarm level database. The alarm level database pre-stores a lookup table of different early warning information and their corresponding alarm levels, formed by operators recording the corresponding alarm levels for different early warning information sequentially. Based on different early warning mechanisms, corresponding responses are executed, and encrypted transmission is performed to achieve a secure and reliable multi-level emergency response, ensuring that alarm information is tamper-proof and leak-proof.

[0104] Generating environmental adjustment information based on illumination detection information and temperature parameters includes the following steps:

[0105] S21: When the light intensity in the light detection information is lower than the preset light reference intensity, obtain the rate of change of light occlusion, the shape characteristics of the occluded area and the movement trajectory within the preset time period from the light detection information.

[0106] Illumination intensity refers to the amount of light energy received per unit area, reflecting the strength of the light. Illumination reference intensity refers to a pre-defined normal illumination range. Preset time period refers to a pre-set time interval, such as 30 seconds. Rate of change refers to how quickly the illumination intensity changes due to the movement of the obstruction. Shape features refer to the geometric properties of the obstruction's outline. Movement trajectory refers to the path the obstruction takes within the monitoring area. The shape of the obstruction is extracted from the illumination detection information using image processing algorithms. The movement speed and path of the obstruction are calculated using target tracking methods. Image processing algorithms and target tracking methods are existing technologies and will not be elaborated upon here. By acquiring the rate of change, the shape features of the obstructed area, and the movement trajectory, the dynamic characteristics of the obstruction are accurately identified, providing data support for subsequent processing.

[0107] S22: Establish a dynamic occlusion model based on the rate of change, shape features, and movement trajectory.

[0108] A dynamic occlusion model is a mathematical model that describes the motion of an occluding object. The rate of change, shape characteristics, and trajectory are input into a time-series analysis model, and the output is a motion model of the occluding object. This is existing technology and will not be elaborated upon here. Building a model from raw data facilitates the determination of whether occlusion is caused by human intervention.

[0109] S23: Determine whether the dynamic occlusion model conforms to the preset artificial dynamic occlusion model.

[0110] The artificial dynamic occlusion model refers to a model that conforms to the characteristics of artificial occlusion, and this model is pre-set. A similarity comparison algorithm is used to compare the dynamic occlusion model with the artificial dynamic occlusion model. If the similarity exceeds a set similarity value, it is judged to conform, thereby reducing false positives and preventing further harm. The similarity comparison algorithm is existing technology and will not be elaborated upon here.

[0111] S24: If the conditions are met, the preset UAV is controlled to perform airspace replacement and collect modal replacement data based on the movement trajectory and shape characteristics. At the same time, the preset acquisition device is controlled to perform reverse displacement compensation based on the movement trajectory, and the collected data is used as compensation data.

[0112] Unmanned aerial vehicles (UAVs) are devices that can take aerial photographs without human piloting. Airspace coverage refers to filling monitoring blind spots. The UAV's flight path is planned based on the movement trajectory and shape characteristics of the obstructed area to ensure coverage. The UAV then flies along this path, using its onboard image detection device to collect high-altitude aerial image data, including environmental images. The collected data is transmitted to the monitoring center for preprocessing to improve quality. Finally, through spatiotemporal alignment and image fusion techniques, the UAV data is integrated with ground monitoring data to fill the monitoring blind spots. Modal coverage data refers to supplementary high-altitude aerial image data collected by the UAV. Reverse displacement compensation refers to adjusting the angle or position of the image detection device on the acquisition equipment in the opposite direction to the movement of the obstructing object to track the target. Compensation data refers to supplementary data, such as visible light images, infrared images, location information, and video data collected by the UAV. This data is combined with ground equipment data to form a complete monitoring dataset. If human-caused obstruction is detected, a UAV takeoff command is triggered to supplement the data in the obstructed area. Through adjustments by the UAV and dynamic equipment, monitoring blind spots caused by human-caused obstruction are eliminated, ensuring data continuity and enabling better tracking of obstructing individuals. Spatiotemporal alignment and image fusion techniques are existing technologies and will not be elaborated here.

[0113] S25: Generate spatiotemporal correlation data based on modal complementation data and compensation data.

[0114] Spatiotemporal correlated data refers to data obtained by time-synchronizing and matching aerial images from a UAV with images from the image detection device on the acquisition equipment. Based on the information collected by the UAV and ground equipment, a fusion algorithm is used to generate multi-angle monitoring results. The fusion algorithm is existing technology and will not be elaborated upon here. By constructing a global monitoring view, the target tracking capability is enhanced.

[0115] S26: Generate spatiotemporal adjustment information based on spatiotemporal correlation data and temperature parameters, and use the spatiotemporal adjustment information as environmental adjustment information.

[0116] Spatiotemporal adjustment information refers to adjustment strategies generated based on spatiotemporal correlation data and temperature parameters, used to guide environmental control equipment in precise adjustments. Different combinations of spatiotemporal correlation data and temperature parameters correspond to different spatiotemporal adjustment information. This information is obtained by querying a pre-set spatiotemporal adjustment information database. The database pre-stores a lookup table of different combinations of spatiotemporal correlation data and temperature parameters with their corresponding adjustment information. The database is formed by operators sequentially testing and measuring the spatiotemporal adjustment information generated from different combinations of spatiotemporal correlation data and temperature parameters, and then recording the results. By adjusting environmental information, resource scheduling is optimized.

[0117] S27: If not, obstacle avoidance information is generated based on the dynamic occlusion model.

[0118] Obstacle avoidance information refers to instructions for avoiding natural occlusion, such as adjusting the angle of the image detection device on the acquisition equipment to avoid areas where fallen leaves are located. If it does not conform to human-caused occlusion, it indicates natural occlusion, which can be avoided through proactive response measures. When the dynamic occlusion model detects a moving target and determines it to be natural occlusion, it uses a coordinate transformation algorithm to locate the position parameters of the occlusion area in the field of view of the image detection device on the acquisition equipment, adjusts the angle range, and then calculates the horizontal / vertical rotation angle and priority. This generates control instructions containing the target avoidance coordinates, rotation direction, and angle. Simultaneously, edge computing is used to verify the adjusted blind spot coverage. Finally, a standardized obstacle avoidance protocol is output to drive the device to execute, and three sets of data, including timestamp, occlusion type, and adjustment parameters, are recorded in the log. By proactively avoiding natural occlusion, the device's response to non-threatening occlusion is reduced, lowering the false alarm rate.

[0119] S28: Generate environmental fusion information based on obstacle avoidance information and temperature parameters, and use the environmental fusion information as environmental adjustment information.

[0120] Environmental fusion information refers to control information that integrates obstacle avoidance information and temperature parameters. Different combinations of obstacle avoidance information and temperature parameters correspond to different environmental fusion information. This fusion information is obtained by querying a pre-set environmental fusion information database. The database pre-stores a lookup table of different combinations of obstacle avoidance information and temperature parameters with their corresponding environmental fusion information. The database is formed by operators sequentially recording the environmental fusion information generated from different combinations of obstacle avoidance information and temperature parameters. This enables intelligent monitoring that adapts to environmental changes, balancing safety and efficiency.

[0121] Generating environmental adjustment information based on illumination detection information and temperature parameters also includes the following steps:

[0122] S291: When the light intensity of the light detection information is higher than the preset light reference intensity, the area with the higher light reference intensity is defined as the highlight area, and at the same time, it is determined whether the temperature parameter is higher than the preset temperature reference parameter.

[0123] Highlight areas refer to regions where the current light intensity exceeds a preset light reference intensity. Temperature reference parameters are standard temperature values ​​set for the current environment, used to determine if the ambient temperature is too high; these are preset. By determining whether the temperature parameter under high light conditions exceeds the temperature reference parameter, the normality of the current natural environment can be judged. Identifying areas of excessive light in the current environment and simultaneously making a preliminary judgment on the temperature provides a foundation for subsequent environmental adjustments that comprehensively consider both light and temperature factors.

[0124] S292: If the temperature is higher than the preset temperature reference parameter, the area above the preset temperature reference parameter is defined as the high temperature zone.

[0125] The high-temperature zone refers to the area where the temperature exceeds the temperature reference parameter. Accurately identifying the areas with excessively high temperatures allows for targeted temperature adjustments to avoid the negative impact of high temperatures on the equipment.

[0126] S293: Generates a high-gloss thermal zone based on the high-gloss and high-temperature zones.

[0127] The high-light-temperature zone refers to the area jointly formed by the high-light and high-temperature zones, that is, a region that simultaneously possesses both high light and high temperature. Based on the previously defined high-light and high-temperature zones and a preset spatial coordinate system, the range and location information of both are determined, and the intersection area is analyzed to obtain the high-light-temperature zone. Identifying the areas significantly affected by light and temperature helps in the targeted adjustment of environmental parameters.

[0128] S294: Obtain the current preset position of the light shield as the current position of the baffle.

[0129] A light-shielding plate is a pre-installed plate used to block light, mounted above the image detection device on the acquisition equipment. The current position of the light-shielding plate refers to its current angular position. The current position data of the light-shielding plate is obtained by reading the angle data from the angle sensor pre-installed on the plate and analyzing it. This data is defined as the current position of the light-shielding plate. Knowing the current position of the light-shielding plate provides data support for subsequent adjustments.

[0130] S295: Calculate the rotation angle of the light-shielding plate based on the current position of the baffle and the high light and heat zone, and use it as the adjustment angle.

[0131] The rotation angle refers to the angle by which the light-shielding panel rotates from its current position to effectively block the high-light-heat area. Based on the current blocking range of the light-shielding panel and the range of the high-light-heat area, the specific angle that the light-shielding panel needs to rotate is determined using geometric calculations and mathematical methods, preparing for subsequent adjustments. Utilizing the position and range of the high-light-heat area and the current preset position of the light-shielding panel, the specific angle that the light-shielding panel needs to rotate is determined using geometric calculations, trigonometric function operations, and other mathematical methods to accurately block the high-light-heat area. For example, in a two-dimensional coordinate system, based on the coordinates of the center position of the high-light-heat area and the current position coordinates of the light-shielding panel, combined with parameters such as the light-shielding direction and effective light-shielding range, the angle that the light-shielding panel needs to rotate is calculated using angle calculation formulas in coordinate geometry, and this calculation result is used as the adjustment angle.

[0132] S296: Generate device adjustment information based on the adjustment angle.

[0133] Equipment adjustment information refers to the specific instructions and parameter information for adjusting the angle of the light shield. Different adjustment angles correspond to different equipment adjustment information, which is obtained by querying a preset equipment adjustment information database. The equipment adjustment information database pre-stores a lookup table of different adjustment angles and their corresponding equipment adjustment information. The database is formed by operators sequentially recording the equipment adjustment information generated for different adjustment angles.

[0134] S297: Based on the equipment adjustment information, the preset light shield is executed, and the light adjustment detection information and temperature adjustment parameters are obtained after the execution.

[0135] Light intensity adjustment detection information refers to the light intensity after the light-shielding panel performs its shading operation. This information, obtained through a light intensity sensor, is used to evaluate the shading effect. For example, if the light intensity in the area measured by the light intensity sensor decreases from 10,000 lux to 500 lux after the light-shielding panel rotates a certain angle, then this 500 lux is the light intensity adjustment detection information. Temperature adjustment parameters refer to the temperature after the light-shielding panel's operation. This information, obtained by re-measuring the ambient temperature through a temperature sensor, reflects the impact of shading on temperature. For example, if the temperature sensor measures a decrease from 35°C to 29°C after shading, this 29°C is the temperature adjustment parameter. This provides feedback data for further evaluation of the shading effect and for environmental control.

[0136] S298: Based on the light regulation detection information and temperature regulation parameters, an environmental regulation factor is generated, and the environmental regulation factor is used as environmental adjustment information.

[0137] Environmental regulation factors refer to adjustment information used to regulate environmental parameters. Different combinations of light intensity regulation detection information and temperature regulation parameters correspond to different environmental regulation factors. The correspondence between environmental regulation factors is determined by querying a pre-set environmental regulation factor database. The environmental regulation factor database pre-stores a lookup table of different combinations of light intensity regulation detection information and temperature regulation parameters and their corresponding environmental regulation factors. The database is formed by operators sequentially recording the environmental regulation factors generated by different combinations of light intensity regulation detection information and temperature regulation parameters.

[0138] S299: If the temperature is not higher than the preset temperature reference parameter, then retrieve the sound data from the multimodal data.

[0139] Sound data refers to ambient sound, acquired through sound sensors such as microphones. When the temperature parameter is not higher than the preset temperature reference parameter, sound data is retrieved from multimodal data to prepare for subsequent environmental adjustments based on sound and other factors.

[0140] S29A: Optimize image acquisition parameters based on illumination detection information.

[0141] Image acquisition parameters refer to settings such as exposure time, sensitivity, white balance, and dynamic range of the light intensity sensor that affect image quality. In security systems, optimized values ​​for image acquisition parameters are calculated based on real-time lighting data. In low-light environments, a combination of parameters—extending exposure time, activating infrared illuminators, and controlling digital gain—is calculated to reduce noise and increase image brightness. In strong-light scenarios, wide dynamic range parameters are calculated to balance contrast between bright and dark areas, while color temperature parameters are adjusted to prevent overexposure. In cases of sudden light sources, anti-flicker algorithm parameters are calculated and combined with local optimization parameters for AI-based identification of key areas to ensure optimal detail clarity and color accuracy in the monitored image, thereby improving the detection rate of abnormal behavior. A low-light environment refers to an environment where the light intensity is lower than a preset light reference intensity.

[0142] S29B: Generate audio-visual adjustment information based on image acquisition parameters and sound data, and use the audio-visual adjustment information as environmental adjustment information.

[0143] Audio-visual adjustment information refers to information generated after comprehensively considering image acquisition information and sound data, used to adjust related equipment. Based on the acquired image acquisition parameters and sound data, sound characteristics are first analyzed to extract abnormal frequency band energy distribution and dynamic volume change trends. Then, directional enhancement parameters are generated using a sound source localization algorithm. Simultaneously, brightness compensation values ​​and exposure corrections in the image parameters are feature-encoded. A temporal correlation model is used to align the audio-visual data streams. Finally, the sound source localization parameters and image optical compensation are fused to output an audio-visual collaborative control command containing spatial enhancement weights and frequency band optimization values. This achieves collaborative optimization of audio and video acquisition and processing, ensuring synchronization and coordination of audio and video during environmental monitoring. This improves the overall effect and user experience of the security monitoring system, enabling monitoring personnel to more accurately judge environmental conditions through audio-visual information. The sound source localization algorithm is existing technology and will not be elaborated upon here.

[0144] Generating device adjustment information based on the adjustment angle includes the following steps:

[0145] S2961: Obtain the monitoring area.

[0146] The monitoring area refers to the range that the image detection device can currently cover and effectively monitor. By reading the parameter settings of the image detection device, including orientation angle and focus, the actual area that can be monitored is obtained. Knowing the monitoring area provides data support for subsequent comparisons with preset monitoring areas, identifying potential monitoring blind spots.

[0147] S2962: When the monitoring area is inconsistent with the preset monitoring area, the unmonitored area shall be named the monitoring blind zone.

[0148] The preset monitoring area refers to the pre-defined range of image acquisition, determined based on the needs of the security system and the scene layout. A monitoring blind spot refers to an area that cannot be monitored. By comparing the currently acquired monitoring area with the preset monitoring area in terms of shape, size, and location, any areas found that cannot cover the preset monitoring area are designated as monitoring blind spots.

[0149] S2963: The coordinate range of the monitoring blind zone is obtained based on the preset spatial coordinate system and used as the blind zone coordinates.

[0150] A spatial coordinate system is a reference coordinate system established to determine the position of an object or region in three-dimensional space. It typically uses a fixed point as the origin and sets three mutually perpendicular coordinate axes, such as the X, Y, and Z axes. The coordinate values ​​uniquely determine the position of any point in space. Blind zone coordinates refer to the area covered by the coordinate position of a monitoring blind zone in the spatial coordinate system.

[0151] Based on the preset spatial coordinate system, the location information of the monitoring blind zone is converted into the coordinate range under this coordinate system to obtain the blind zone coordinates. This provides accurate geometric location parameters for subsequent calculation of the difference in the monitoring area, which helps to accurately determine the size and location of the monitoring blind zone.

[0152] S2964: Calculate the difference in the monitoring area based on the blind zone coordinates.

[0153] The monitoring area difference refers to the degree of difference between the preset monitoring area and the current actual monitoring area. Based on the blind zone coordinates and combined with the coordinate range of the preset monitoring area, a geometric calculation method is used to determine the area deviation between the two and use it as the monitoring area difference. By quantitatively reflecting the degree of difference between the current monitoring area and the preset target area, specific data support is provided for subsequent analysis.

[0154] S2965: Obtain the rotation angle difference and elevation angle difference based on the difference in the monitoring area and control the image detection device to perform the operation.

[0155] By adjusting the horizontal rotation angle and vertical elevation angle, the observation direction of the equipment is dynamically adjusted to align with the target location or abnormal area within the monitoring zone. Rotation angle difference refers to the angular difference in rotation of the image detection device on the horizontal plane. Elevation angle difference refers to the angular difference in elevation or depression of the monitoring equipment on the vertical plane. Based on the differences in the monitoring area, the required rotation and elevation angle differences for the monitoring equipment are calculated using trigonometric functions and other methods. This determines the specific angle parameters that the monitoring equipment needs to adjust, enabling the equipment to perform corresponding actions, expanding the monitoring area, reducing blind spots, and improving the comprehensiveness and effectiveness of monitoring.

[0156] S2966: Generate focus parameter corrections based on rotational angular difference, pitch angular difference, and blind zone coordinates.

[0157] Focus parameter correction refers to the focusing parameters that enable the equipment to image the monitoring blind area more clearly. Different combinations of rotational angular difference, elevation angular difference, and blind area coordinates correspond to different focus parameter corrections. These correction values ​​are obtained by querying a pre-set focus parameter correction database. This database contains a lookup table of different combinations of rotational angular difference, elevation angular difference, and blind area coordinates with their corresponding focus parameter corrections. The database is created by operators through sequential experimental measurements of the focus parameter corrections generated by different combinations of rotational angular difference, elevation angular difference, and blind area coordinates. Based on the adjusted attitude of the monitoring equipment and the location characteristics of the monitoring blind area, the focus parameters are precisely adjusted to improve the image acquisition quality of the monitoring blind area.

[0158] S2967: Generate equipment adjustment information based on focus parameter correction, rotation angle difference, and pitch angle difference.

[0159] Equipment adjustment information refers to the instructions for adjusting the rotation angle of the image detection device, including parameters such as rotation angle difference, elevation angle difference, and focus parameter correction. The focus parameter correction, rotation angle difference, and elevation angle difference are integrated and generated into complete equipment adjustment information according to a specific data format and communication protocol. For example, the rotation angle difference, elevation angle difference, and focus parameter correction are encoded into specific instruction codes or numerical parameters, combined into a data frame or instruction packet, and sent as equipment adjustment information to the control system of the monitoring equipment. Through comprehensive and precise equipment adjustment instructions, the monitoring equipment can be guided to simultaneously adjust its direction and focus parameters, achieving effective coverage of blind spots and high-quality image acquisition.

[0160] The method for determining the presence of abnormal behavior includes the following steps:

[0161] S61: Extract behavioral features from multimodal data.

[0162] Behavioral features refer to the salient attributes or key information of a target object's behavioral patterns. Common behavioral features include the target object's movement trajectory, such as movement speed, direction, and path shape; posture changes, such as the angles of different body parts; action frequency, such as the number of times a certain action occurs per unit of time; and dwell time. Extracting behavioral features provides the foundational data for subsequent behavioral analysis.

[0163] S62: When a behavioral feature does not conform to the preset behavioral baseline model, extract items that do not conform to the behavioral baseline model from the behavioral features and mark them as abnormal items.

[0164] A behavioral baseline model refers to a pre-defined behavioral model of normal behavior. Abnormal items refer to behavioral characteristics that do not conform to the behavioral baseline model.

[0165] S63: Extract the number of abnormal items and use it as the number of abnormal items.

[0166] The number of anomalies refers to the number of abnormal items in the behavioral characteristics. It provides a simple and intuitive quantitative indicator to measure the degree of behavioral abnormality, and provides basic data support for further assessment of the risk level of abnormal behavior and determination of whether an alarm is needed.

[0167] S64: Generate anomaly confidence based on the number of abnormal items and anomalies.

[0168] Anomaly confidence score reflects the degree of certainty that current behavior is considered abnormal, typically a value between 0 and 1. A higher value indicates a higher probability of abnormal behavior; a lower value indicates a higher probability of normal behavior. Using a pre-established confidence score calculation model, such as logistic regression, the anomaly items and their number are used as input features to calculate the anomaly confidence score. Logistic regression is an existing technology and will not be elaborated upon here. By comprehensively considering the types and numbers of anomaly items, a more comprehensive quantitative indicator reflecting the probability of anomalies is generated, making subsequent judgments on abnormal behavior more scientific and reasonable, and avoiding misjudgments that may result from relying on a single factor.

[0169] S65: Determine whether the anomaly confidence level exceeds the preset anomaly baseline confidence level.

[0170] The anomaly baseline confidence level is a pre-set value used to determine whether behavior is abnormal. By judging whether the pre-set anomaly baseline confidence level is exceeded, it is determined whether an alarm is needed, ensuring that the system can issue alarms reasonably according to actual security needs and avoid excessive false alarms or missed alarms.

[0171] S66: If it exceeds the limit, it is determined that there is abnormal behavior.

[0172] When the judgment result is that the confidence level of the abnormality exceeds the preset value, the current behavior is directly judged as abnormal behavior.

[0173] S67: If the number of occurrences of abnormal items is not exceeded, the number of occurrences of abnormal items will be retrieved from the behavioral characteristics as the number of abnormalities.

[0174] The anomaly count refers to the total number of times an anomaly item occurs. If the anomaly confidence level does not exceed the preset value, further analysis is performed by obtaining the anomaly count to avoid ignoring potentially recurring anomalies due to a single, accidental anomaly.

[0175] S68: If the number of anomalies exceeds the preset number, a second verification will be performed based on multimodal data.

[0176] The preset number of occurrences refers to a pre-defined number of times an abnormal item will appear, such as 5 times, used to determine whether further verification is needed. Secondary verification involves analyzing and confirming the current behavior again using comprehensive multimodal data to determine if any abnormal behavior exists. When the number of abnormal occurrences exceeds the preset number, the secondary verification process is initiated to perform a more in-depth analysis of the multimodal data. Specific analysis steps are detailed in S681 to S685.

[0177] Secondary validation based on multimodal data includes the following steps:

[0178] S681: Retrieves audio data from multimodal data.

[0179] Audio data is retrieved from multimodal data. This retrieved audio data provides data support for the subsequent generation of acoustic texture feature data.

[0180] S682: Generate acoustic texture feature data based on sound data and behavioral characteristics.

[0181] Vocal signature data refers to comprehensive data that integrates voice data and behavioral features, used to more fully describe the behavior of a target object. Integrating voice data and behavioral features using feature fusion algorithms is existing technology and will not be elaborated upon here. By fusing voice and behavioral features, normal and abnormal behaviors can be easily and accurately identified and distinguished, improving the accuracy and reliability of behavior analysis.

[0182] S683: Match the acoustic texture feature data with the preset acoustic texture database and extract the one with the highest matching degree as the initial anomaly judgment value.

[0183] A voiceprint database refers to a pre-set database that stores a large number of known voiceprint feature data samples, including feature data of both normal and abnormal behaviors. The initial anomaly judgment value is the matching value found in the voiceprint database corresponding to the sample with the highest matching degree to the current voiceprint feature data. It represents the similarity between the current feature data and known behavioral patterns and can be used to initially determine whether the behavior is abnormal. Using pattern matching algorithms to match the current voiceprint feature data one by one with all samples in the voiceprint database is existing technology and will not be elaborated upon here. Utilizing known voiceprint feature data as a reference narrows the scope of analysis and reduces the workload of subsequent further verification.

[0184] S684: Calculate the difference between the initial anomaly judgment value and the baseline matching degree as the anomaly difference.

[0185] The baseline match rate refers to the ideal match rate standard, which is preset. The anomaly difference refers to the degree of difference between the initial anomaly judgment value and the baseline match rate, used to quantify the degree of deviation between the current behavior and the normal behavior pattern. The anomaly difference is calculated by subtracting the anomaly judgment value from the baseline match rate, providing a quantitative indicator to measure the degree of deviation between the current behavior and the normal behavior pattern, and providing a more specific basis for subsequent judgment on whether it is an abnormal behavior.

[0186] S685: When the abnormal difference exceeds the preset abnormal difference range, it is determined that there is abnormal behavior.

[0187] The abnormal difference range is a pre-set range used to determine whether an abnormal difference exceeds the normal range. If it exceeds the abnormal difference range, it is determined that there is abnormal behavior, providing a clear judgment result for security early warning and facilitating accurate intervention.

[0188] The method for determining the presence of abnormal behavior also includes the following steps:

[0189] S71: Obtain the current pedestrian density from multimodal data.

[0190] Current crowd density refers to the number of people within the monitored area at the current moment. This is calculated by identifying the number of features matching preset personnel characteristics from multimodal data, and then proportionally calculating this number to the area corresponding to the preset monitoring area. Personnel characteristics refer to the features of various parts of a person's body, which are obtained through pre-input. The area corresponding to the monitoring area is also obtained through pre-input.

[0191] Real-time statistics on the number of people in the monitored area provide basic data support for subsequent judgment of crowd density and analysis of group behavior, ensuring that the system can understand the distribution of people in a timely manner.

[0192] S72: Determine whether the current crowd density exceeds the preset density value.

[0193] The density threshold refers to the limit for crowd density, which is predetermined based on safety standards, venue capacity, and historical experience. By determining whether the current crowd density exceeds the preset density threshold, situations of excessive crowding can be detected in a timely manner, providing a basis for subsequent crowd control measures or enhanced monitoring.

[0194] S73: If the density value is exceeded, the movement trajectory, movement speed and distribution characteristics of the crowd will be collected in real time.

[0195] Movement trajectory refers to the movement path of a group of people or individuals over time. Movement speed refers to the speed at which a group of people or individuals move. Distribution characteristics refer to the spatial distribution of a group of people, including the degree of dispersion and the uniformity of distribution. By collecting these parameters through image detection devices, detailed dynamic information about the crowd can be obtained in a timely manner in densely populated areas. This provides data support for subsequent group behavior analysis and abnormal individual identification, and helps to detect potential dangerous behaviors or abnormal dynamics in advance.

[0196] S74: Generate dynamic data of the population based on movement trajectory, movement speed and distribution characteristics.

[0197] Crowd dynamic data refers to a comprehensive dataset reflecting the overall movement and distribution changes of a crowd, including information such as average movement speed, main movement direction, changes in clustering areas, and dispersal trends. It describes the dynamic evolution of a crowd over time. The process involves comprehensive analysis and processing of real-time collected movement trajectories, speeds, and distribution characteristics. First, the average movement speed and standard deviation of the crowd are calculated to determine the main movement direction. Then, the changes in clustering areas over time are analyzed, for example, by using clustering algorithms to detect changes in the displacement of cluster centers and the degree of clustering—existing technology, which will not be elaborated here. Finally, this information is integrated into a structured database, including fields such as timestamp, average speed, main direction, clustering area coordinates, and clustering degree, generating crowd dynamic data. This is implemented using a custom data structure. The generated comprehensive data reflects the overall movement state of the crowd, providing concise and comprehensive data input for subsequent crowd dynamic model building, facilitating further analysis and modeling.

[0198] S75: Establish a crowd dynamic model based on crowd dynamic data and current crowd density, and label individual behavioral characteristics.

[0199] A crowd dynamics model is a model used to describe and predict the dynamic behavior of a crowd under different conditions. Individual behavioral characteristics refer to the specific behavioral attributes exhibited by each individual in the crowd, such as sudden changes in walking direction, abnormal acceleration or deceleration in speed, or deviation from the mainstream direction of the crowd. The specific steps for establishing a crowd dynamics model are described in sections S751 to S757. By establishing a crowd dynamics model, the overall behavior of the crowd can be simulated and predicted, while individual behavioral characteristics are labeled. First, a crowd dynamics model is constructed using crowd dynamics data and current crowd density, covering the overall dynamic behavior of the crowd. Then, for each individual in the crowd, their behavior is monitored in real time. Based on set characteristic criteria, such as whether there is a sudden change in walking direction, abnormal acceleration or deceleration in speed, or deviation from the mainstream direction of the crowd, it is determined whether the individual possesses these specific abnormal behavioral characteristics. Once an individual is determined to have a certain characteristic, it is recorded and identified using a specific labeling method, such as marking it with a specific color or symbol on the model visualization interface. This helps to accurately identify individuals with abnormal behavior within the crowd, providing strong support for subsequent steps.

[0200] S76: Generate vocal data based on individual behavioral characteristics and voice data.

[0201] Voice-based behavioral data refers to a comprehensive data type that integrates individual behavioral and vocal features to more fully describe an individual's behavioral state. This involves fusing labeled individual behavioral feature data with vocal data. First, the vocal data undergoes preprocessing, such as noise removal, extraction of specific sounds, and identification of specific vocal events related to the individual—existing technologies that will not be elaborated upon here. Then, the individual behavioral features are combined with corresponding vocal features to form the voice-based behavioral data. This can be achieved through data association algorithms, which are existing technologies and will not be elaborated upon here. For example, based on timestamps and spatial location information, individual vocal events are matched with behavioral features to generate a voice-based behavioral data structure containing fields such as individual identifier, behavioral feature vector, and vocal feature vector. By fusing individual behavioral and vocal features, richer data describing the individual's state is generated, which helps to more accurately identify abnormal individual behavior, avoids misjudgments caused by relying on a single feature, and improves the accuracy and reliability of abnormal behavior detection.

[0202] S77: If the voice data does not match the preset voice baseline data within a preset time period, then mark the abnormal individual and determine that there is abnormal behavior.

[0203] The preset time period refers to a pre-defined time interval, such as 30 seconds before and 1 minute after the event. The voice behavior baseline data refers to voice behavior data representing normal behavior, and is also pre-set. An abnormal individual refers to an individual in a group whose voice behavior data differs significantly from the voice behavior baseline data. Their behavior may pose a security threat or require special attention, such as an individual suddenly and rapidly walking against the flow of traffic in a crowd while shouting. By comparing voice behavior data that does not match the preset time period with the preset voice behavior baseline data, abnormal individuals are identified, providing timely warnings of abnormal behavior for security monitoring and enhancing the detection capability and response speed of abnormal behavior in complex scenarios.

[0204] Establishing a crowd dynamics model based on crowd dynamics data and current crowd density includes the following steps:

[0205] S751: Generate an initial dynamic model based on population dynamic data.

[0206] The initial dynamic model refers to a preliminary model constructed based on crowd dynamic data, used to describe information such as the average movement speed and main direction of movement of the crowd. The crowd dynamic data is input into a pre-defined model architecture to generate the initial dynamic model. By continuously inputting new crowd dynamic data, incremental learning is used to dynamically correct the model parameters, enabling online learning and updates. This provides a foundation for further analysis and adjustments, ensuring the system can respond promptly to changes in crowd dynamics. The model architecture can be based on continuity equations and motion equations from fluid mechanics. Generating models based on this architecture is existing technology and will not be elaborated further.

[0207] S752: Determine whether the initial dynamic model matches the baseline model of group behavior.

[0208] A group behavior baseline model is a pre-established reference model that represents normal group behavior patterns. It is typically built based on a large amount of historical data and behavioral characteristics under normal scenarios. By judging whether the initial dynamic model matches the group behavior baseline model, it is possible to effectively identify whether the initial model accurately reflects normal group behavior, thereby determining whether further adjustments are needed to improve the model's accuracy and reliability.

[0209] S753: If they match, the initial dynamic model is adjusted based on the current crowd density to form a group dynamic model.

[0210] When the initial dynamic model matches the baseline model of crowd behavior, the current pedestrian flow is added to the initial dynamic model for parameter adjustment to obtain the final crowd dynamic model. This improves the model's adaptability to the current crowd state, enabling it to more accurately predict and describe crowd behavior and provide more reliable data support for security decisions.

[0211] S754: If they do not match, extract the abnormal region from the initial dynamic model.

[0212] Anomaly regions refer to areas in the initial dynamic model that differ significantly from the baseline model. If the initial dynamic model does not match the baseline model of group behavior, it is necessary to extract anomaly regions from the initial dynamic model. First, the baseline model of group behavior needs to be determined, representing the normal behavior pattern of a crowd. Then, the difference index between the initial dynamic model and the baseline model is calculated. A range is set; when the difference index exceeds the range, anomaly regions are identified. Next, these significantly different parts in the initial model are located and extracted as anomaly regions. The extracted anomaly regions are input into the subsequent risk assessment and response modules to provide a basis for them. For example, in a surveillance video scenario, the initial model describes crowd flow, while the baseline model represents a normal flow pattern. When the difference index between the two exceeds the range, anomaly regions are located and extracted, which may be areas of crowd aggregation or abnormal individual movement. These regions can be identified by calculating the difference index between the models. Extracting anomaly regions helps to promptly detect abnormal behaviors or distributions within a crowd, providing a basis for subsequent risk assessment and response measures.

[0213] S755: Calculate the rate of change of the area of ​​the abnormal region within a preset time period and generate the diffusion rate of the abnormal region.

[0214] The preset time period refers to a pre-determined time interval based on the actual application scenario, used to monitor changes in the area of ​​abnormal regions. The area change rate refers to the speed at which the area of ​​the abnormal region changes within the preset time period. The anomaly spread rate refers to the speed at which the abnormal region expands within the preset time period, derived from the area change corresponding to the abnormal region within that time period. The ratio of area change to time is then calculated to obtain the area change rate, and the anomaly spread rate is deduced from this rate. Different area change rates correspond to different anomaly spread rates, which are obtained by querying a preset anomaly spread database. The anomaly spread database pre-stores a table mapping different area changes to their corresponding anomaly spread rates. This database is formed through continuous dynamic monitoring by researchers of changes in the anomaly spread rate caused by different area changes. This provides data support for taking proactive measures and helps prevent potential security risks.

[0215] S756: Calculate the group risk index based on the diffusion speed of the abnormal area and the current population density.

[0216] The group risk index is an indicator that comprehensively reflects the risk level of the current population state, taking into account the spread rate of abnormal areas and the current population density. Data on the spread rate of abnormal areas and the current population flow are collected. Using a risk assessment model, these two parameters are used as input variables to calculate the group risk index, which helps to quickly identify high-risk scenarios and take appropriate security measures in a timely manner. The risk assessment model is existing technology and will not be elaborated upon here.

[0217] S757: Establish a multi-level population model based on the initial dynamic model, abnormal regions, and risk indices, and use the multi-level population model as the population dynamic model.

[0218] A multi-layered population model is a comprehensive model comprising multiple levels. It utilizes information from anomalous regions to construct an anomaly layer, representing areas of abnormal behavior or distribution. A risk layer is constructed by combining risk indices, reflecting the risk levels in different regions. These layers are then integrated with an initial dynamic model or other relevant data to form a multi-layered model structure. Ultimately, using this multi-layered population model as a population dynamic model allows for a more comprehensive and accurate description and prediction of population dynamics and risk profiles.

[0219] After the early warning information is encrypted and transmitted to the security system terminal for early warning purposes, the following steps are also included:

[0220] S51: Generate a revised risk index based on environmental adjustment information and multimodal data.

[0221] The revised risk index is the result of correcting the original risk index, taking into account environmental adjustment information and multimodal data, to more accurately reflect the risk level of the current environment and behavior. Using a risk assessment model, the revised risk index is calculated with environmental adjustment information and multimodal data characteristics as input variables. It comprehensively considers the impact of environmental factors and population behavior factors on risk. Input variables include environmentally adjusted light intensity and temperature, as well as population density, movement speed, and the number of abnormal behaviors from the multimodal data. The model output is the revised risk index, reflecting the overall risk level under the current environmental and population conditions.

[0222] S52: Generate the rate of increase based on the modified risk index.

[0223] The rate of increase refers to the speed at which the modified risk index changes within a unit of time, used to measure the trend of change in risk level. It is calculated by taking the rate of change of the modified risk index within a unit of time. For example, if the unit of time is 15 seconds, the modified risk index is recorded at each time point within those 15 seconds. Then, the difference in the index between adjacent time points is calculated, and the average rate of change is obtained; this is the rate of increase. Calculating the rate of increase provides a quantitative indicator to measure the trend of change in risk level, helping to promptly detect rapid increases in risk and providing a basis for subsequent emergency measures.

[0224] S53: When the ascent rate is greater than the preset ascent reference rate, the difference between the ascent rate and the preset ascent reference rate is calculated and used as the ascent rate difference.

[0225] The baseline rise rate refers to the rise rate threshold, used to determine whether the rate of increase of the corrected risk index is abnormal; it is preset. The rise rate difference measures the degree of abnormality of the current rise rate. The current rise rate is compared with the preset baseline rise rate. If the current rise rate is greater than the preset value, the difference between the two is calculated and stored as the rise rate difference, providing a basis for subsequent calls to backup monitoring equipment.

[0226] S54: Generate backup monitoring retrieval information based on the difference in the rate of rise, and call the preset backup monitoring equipment for linkage based on the backup monitoring retrieval information.

[0227] Backup monitoring retrieval information refers to the retrieval commands and parameters of backup monitoring devices. Backup monitoring devices are pre-configured to provide additional monitoring capabilities when the primary monitoring equipment is insufficient or requires enhanced monitoring. The urgency of backup monitoring retrieval is determined based on the magnitude of the rate of increase difference. Devices requiring retrieval are then selected from a pre-set list of backup monitoring devices based on their urgency. The backup monitoring device list details the identifier, location, monitoring range, device type, retrieval priority, and applicable scenarios for all backup devices. The backup monitoring retrieval information is then used to link with the pre-set backup monitoring devices, thereby improving monitoring accuracy.

[0228] S55: Generate early warning feedback information based on the revised risk index and the rate of increase.

[0229] Early warning feedback information refers to comprehensive information including the modified risk index and the rate of increase, used to provide security personnel with detailed feedback on the current risk situation. Different combinations of modified risk indices and rates of increase correspond to different early warning feedback messages. The early warning feedback information is obtained by querying a pre-set early warning feedback information database. This database pre-stores a lookup table of different combinations of modified risk indices and rates of increase and their corresponding early warning feedback messages. The database is formed by operators sequentially recording the early warning feedback messages generated from different combinations of modified risk indices and rates of increase. This helps security personnel quickly understand the current risk situation and provides support for decision-making.

[0230] S56: Encrypt the backup monitoring information and early warning feedback information and update them synchronously to the multi-level early warning strategy library of the security system terminal.

[0231] A multi-level early warning strategy database refers to a database of multi-level early warning strategies stored in the security system terminal, used to take corresponding early warning measures based on different risk levels. Backup monitoring information and early warning feedback information are encrypted, and the video stream is encrypted using the national cryptographic algorithm SM4, with a key dynamic update cycle of ≤5 minutes. This ensures the security of data transmission. Then, the encrypted data is synchronously updated to the multi-level early warning strategy database of the security system terminal to improve the reliability and responsiveness of the security system.

[0232] Based on the same inventive concept, embodiments of the present invention provide an AIoT-based intelligent security early warning system, comprising:

[0233] The acquisition module is used to acquire illumination detection information, temperature parameters, and multimodal data;

[0234] A memory for storing programs such as the aforementioned AIoT-based intelligent security early warning method;

[0235] The processor loads and executes programs from memory.

[0236] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0237] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An AIoT-based intelligent security early warning method, characterized in that, Comprise: S1: real-time acquisition of light detection information and temperature parameters in natural environment; S2: generating environment adjustment information based on the light detection information and the temperature parameters and outputting to the acquisition device to collect multi-modal data; S3: generating behavior information based on the multi-modal data; S4: when the behavior information exists abnormal behavior, then generating early warning information based on the environment adjustment information and the behavior information; S5: based on the early warning information trigger hierarchical early warning mechanism, while the early warning information encrypted transmission to the security system terminal to carry out early warning; Based on the light detection information and the temperature parameters generate environment adjustment information includes: S21: when the light intensity in the light detection information is lower than the preset light reference intensity, the change rate of light shielding in the preset time period, the shape feature of the shielding area and the moving track are obtained from the light detection information; S22: based on the change rate, the shape feature and the moving track to establish a dynamic shielding model; S23: determine whether the dynamic shielding model conforms to the preset human dynamic shielding model; S24: if it is consistent, based on the moving track and the shape feature control the preset unmanned aerial vehicle to carry out airspace compensation collection modal compensation data, based on the moving track control the preset acquisition device to execute reverse displacement compensation, and the collected data as compensation data; S25: based on the modal compensation data and the compensation data generate space-time correlation data; S26: based on the space-time correlation data and the temperature parameters generate space-time adjustment information, and the space-time adjustment information as the environment adjustment information; S27: if it is not consistent, based on the dynamic shielding model to generate obstacle avoidance information; S28: based on the obstacle avoidance information and the temperature parameters generate environment fusion information, and the environment fusion information as the environment adjustment information. 2.The AIoT-based intelligent security early warning method according to claim 1, characterized in that, Based on the light detection information and the temperature parameters generate environment adjustment information also includes: S291: when the light intensity of the light detection information is higher than the preset light reference intensity, then define the area higher than the preset light reference intensity as high light area, and determine whether the temperature parameter is higher than the preset temperature reference parameter; S292: if higher than the preset temperature reference parameter, define the area higher than the preset temperature reference parameter as high temperature area; S293: based on the high light area and the high temperature area to generate high light heat area; S294: obtain the current position of the preset light shield as the current position of the baffle; S295: based on the current position of the baffle and the high light heat area calculate the rotation angle of the light shield as the adjustment angle; S296: based on the adjustment angle to generate device adjustment information; S297: based on the device adjustment information control the preset light shield to execute, and obtain the light adjustment detection information and temperature adjustment parameter after execution; S298: based on the light adjustment detection information and the temperature adjustment parameter generate environment control factor, and the environment control factor as the environment adjustment information; S299: if it is not higher than the preset temperature reference parameter, then retrieve sound data from the multi-modal data; S29A: optimizing image acquisition parameters based on the illumination detection information; S29B: generating sound-picture adjustment information based on the image acquisition parameters and the sound data, and taking the sound-picture adjustment information as the environment adjustment information. 3.The AIoT-based intelligent security early warning method of claim 2, characterized in that, Generating device adjustment information based on the adjustment angle includes: S2961: obtaining a monitoring area; S2962: when the monitoring area is inconsistent with a preset monitoring area, naming an unmonitored area as a monitoring blind area; S2963: obtaining a coordinate range of the monitoring blind area as a blind area coordinate based on a preset spatial coordinate system; S2964: calculating a monitoring area difference value based on the blind area coordinate; S2965: obtaining a rotation angle difference and a tilt angle difference based on the monitoring area difference value and controlling an illumination intensity sensor to execute; S2966: generating a focus parameter correction amount based on the rotation angle difference, the tilt angle difference, and the blind area coordinate; S2967: generating the device adjustment information based on the focus parameter correction amount, the rotation angle difference, and the tilt angle difference. 4.The AIoT-based intelligent security early warning method according to claim 1, characterized in that, The determination method of the abnormal behavior includes: S61: extracting a behavior feature from the multi-modal data; S62: when the behavior feature does not conform to a preset behavior reference model, extracting an item inconsistent with the behavior reference model from the behavior feature and marking it as an abnormal item; S63: extracting the number of abnormal items as an abnormal number; S64: generating an abnormal confidence based on the abnormal item and the abnormal number; S65: determining whether the abnormal confidence exceeds a preset abnormal reference confidence; S66: if it exceeds, determining that the abnormal behavior exists; S67: if it does not exceed, retrieving the number of times the abnormal item appears from the behavior feature as an abnormal number of times; S68: if the abnormal number of times exceeds a preset number of times, performing secondary verification based on the multi-modal data. 5.The AIoT-based intelligent security early warning method according to claim 4, characterized in that, The secondary verification based on the multi-modal data includes: S681: retrieving sound data from the multi-modal data; S682: generating sound-state line feature data based on the sound data and the behavior feature; S683: matching the sound-state line feature data with a preset sound-state line database and extracting the highest matching degree as an abnormal preliminary judgment value; S684: calculating a difference between the abnormal preliminary judgment value and a reference matching degree as an abnormal difference; S685: when the abnormal difference exceeds a preset abnormal difference interval, determining that the abnormal behavior exists.

6. The AIoT-based intelligent security early warning method according to claim 5, characterized in that, The determination method of the abnormal behavior further includes: S71: obtaining a current crowd density from the multi-modal data; S72: determining whether the current crowd density exceeds a preset dense value; S73: if it exceeds the dense value, collecting the moving trajectory, moving speed, and distribution characteristics of the crowd in real time; S74: generating crowd dynamic data based on the moving trajectory, the moving speed, and the distribution characteristics; S75: establishing a crowd dynamic model based on the crowd dynamic data and the current crowd density, and marking individual behavior features; S76: generating sound line data based on the individual behavior features and the sound data; S77: If the sound movement data does not match the preset sound movement reference data within a preset time period, an abnormal individual is marked and the abnormal behavior is determined to exist.

7. The AIoT-based intelligent security early warning method according to claim 6, characterized in that, The population dynamic model is established based on the population dynamic data and the current crowd density, including: S751: An initial dynamic model is generated based on the population dynamic data; S752: It is determined whether the initial dynamic model matches a population behavior reference model; S753: If it matches, the initial dynamic model is adjusted based on the current crowd density to form a population dynamic model; S754: If it does not match, an abnormal area is extracted from the initial dynamic model; S755: The area change rate of the abnormal area within a preset time period is calculated to generate an abnormal area diffusion speed; S756: A population risk index is calculated based on the abnormal area diffusion speed and the current crowd density; S757: A population multi-layer model is established based on the initial dynamic model, the abnormal area, and the risk index, and the population multi-layer model is taken as the population dynamic model. 8.The AIoT-based intelligent security early warning method according to claim 7, characterized in that, After the early warning information is encrypted and transmitted to the security system terminal for early warning, it further includes: S51: A revised risk index is generated based on the environmental adjustment information and the multi-modal data; S52: An upward rate is generated based on the revised risk index; S53: When the upward rate is greater than a preset upward reference rate, the difference between the upward rate and the preset upward reference rate is calculated as an upward rate difference; S54: Backup monitoring call information is generated based on the upward rate difference, and a preset backup monitoring device is called for linkage based on the backup monitoring call information; S55: Early warning feedback information is generated based on the revised risk index and the upward rate; S56: The backup monitoring call information and the early warning feedback information are encrypted and synchronously updated to the multi-level early warning strategy library of the security system terminal.

9. An AIoT-based intelligent security and protection early warning system, characterized in that, It includes: An acquisition module for acquiring illumination detection information, temperature parameters, and multi-modal data; A memory for storing an AIoT-based intelligent security early warning method according to any one of claims 1-8; A processor for loading and executing programs in the memory.

Images