Industrial safety risk grading warning method, device and system based on artificial intelligence dynamic priority, storage medium and program

Abstract

The invention provides an artificial intelligence dynamic priority-based industrial safety risk grading warning method, device and system, a storage medium and a program, and relates to the technical field of risk assessment. S10, static risk area division: determining a static risk area grade L based on factory three-dimensional map data, historical accident data and equipment distribution area data; s20, positioning initialization and database establishment: outputting real-time three-dimensional coordinate data (x, y, z) according to a multi-sensor fusion algorithm, establishing a work ticket risk level mapping library, and mapping a work risk level W in combination with work ticket content; and S30, multi-source data acquisition: generating dynamic risk parameters based on the pre-trained YOLOv8 model. Through cooperation of the above structures, compared with the prior art, the method has the following beneficial effects: dynamic linkage determination of operation risks and regional risks can be realized through real-time fusion of multi-source information such as personnel positioning data, equipment state parameters, regional risk changes and the like.

Description

Technical Field

[0001] The present invention relates to the field of risk assessment technology, and in particular to an industrial safety risk classification warning method, device, system, storage medium and program based on artificial intelligence dynamic priority. Background Art

[0002] At present, the safety risk level management of maintenance personnel in the field of industrial production safety mainly relies on static classification methods. Usually, the work is divided into four fixed levels of "low risk", "general risk", "relatively high risk" and "major risk" according to the content of the work ticket. However, this static classification method has significant limitations: First, the actual risk level of maintenance personnel will be dynamically adjusted as the work status changes. For example, when a worker returns to the office building to rest after completing a high-risk job (such as equipment maintenance), his risk level should be reduced from "relatively high risk" or "major risk" to "low risk", but the existing static management model cannot update this change in real time, resulting in a disconnect between risk assessment and actual scenarios.

[0003] Secondly, the risk level of the on-site operation area may also change dynamically due to factors such as abnormal equipment status, a sudden increase in personnel density, or sudden accidents (for example, equipment failure causes the regional risk level to rise from Level II to Level IV). Traditional methods only divide static areas based on historical data and cannot respond to dynamic risk factors in a timely manner.

[0004] In addition, safety supervisors face two major challenges in their actual work: first, it is difficult to quickly obtain the real-time risk level of the area where the maintenance personnel are currently located, such as the increase in risk level after moving from the ordinary operation area to the high-voltage area; second, it is impossible to synchronously grasp the correlation between personnel location, operation content and regional risk, resulting in delayed supervision decision-making. Therefore, there is an urgent need to establish dynamic safety risk management methods, systems and devices, and realize the dynamic linkage judgment of operation risk and regional risk through real-time integration of multi-source information such as personnel positioning data, equipment status parameters, and regional risk changes, and push real-time risk level, location coordinates and alarm information to the monitoring center, mobile phone and tablet through the 5G network. Summary of the Invention

[0005] In response to the shortcomings of the above-mentioned existing technologies, the technical problem to be solved by the present invention is to provide an industrial safety risk grading warning method, device, system, storage medium and program based on artificial intelligence dynamic priority, which can realize dynamic linkage judgment of operation risk and regional risk through real-time integration of multi-source information such as personnel positioning data, equipment status parameters, regional risk changes, etc., and push real-time risk level, location coordinates and alarm information to the monitoring center, mobile phone and tablet through the 5G network.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is as follows: the present invention provides an industrial safety risk classification warning method based on artificial intelligence dynamic priority, comprising the following steps: S10. Static risk area classification: Determine the static risk area level L based on the plant's three-dimensional map data, historical accident data, and equipment distribution area data; S20, positioning initialization and database establishment: Output real-time three-dimensional coordinate data (x, y, z) based on the multi-sensor fusion algorithm, establish a work ticket risk level mapping library, and map the operation risk level W based on the work ticket content; S30, Multi-source data collection: Generate dynamic risk parameters based on the pre-trained YOLOv8 model; S40, dynamic risk level prediction: integrating the static risk area level L, the operation risk level W and the dynamic risk parameters to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , predicting dynamic risk level ;According to the dynamic risk level Determine the final risk level ; S50, early warning and dynamic display: according to the final risk level Trigger multi-modal warning signals, including LED display, voice prompts and vibration feedback; S60, Emergency response linkage and data tracing: When an abnormal event is detected, the emergency event mechanism is triggered.

[0007] In the preferred solution, the specific steps of step S10 are as follows: S11. Input the factory 3D map data, historical accident GPS coordinates and equipment distribution area data; S12. Construct an accident density heat map G(x,y,z) using the KDE algorithm and assign risk weights to each device type to generate device risk areas. , the formula is as follows: ; Where I is the indicator function, To represent the risk weight of the i-th category equipment, it is a fixed value determined according to the equipment type; S13. Overlay the accident density heat map G(x, y, z) with the equipment risk area R(x, y, z), and use the K-means clustering algorithm combined with the three-dimensional minimum bounding box algorithm to divide the four-level static risk areas, generating a static area electronic map in GeoJSON format containing coordinate boundaries; Among them, the static risk level L∈{I-IV} is determined according to the four-level static risk area. The level I static risk area is a low-risk static area, the level II static risk area is a general risk static area, the level III static risk area is a large risk static area, and the level IV is a high-risk static area.

[0008] In the preferred solution, the specific steps of step S2 are as follows: S21. Multi-sensor fusion positioning: Obtain a safety warning light device that integrates GPS, UWB, and IMU sensors, and output real-time three-dimensional coordinates (x, y, z) through the EKF algorithm. The calculation formula is as follows: Equation of state: ; Among them, x, y, z represent the position coordinates of GPS in three-dimensional space. are the velocities in the x, y, and z directions, respectively. ; is the acceleration in the x, y, and z directions, is the deterministic acceleration component; Observation equation: ; Where H is the sensor observation matrix, is the process noise, is the observation noise; S22. Work Ticket Risk Mapping: Establish a work ticket risk level mapping library, and obtain a set of standard risk feature vectors based on the work ticket risk level mapping library. ; Obtain work ticket content, including operation type O, equipment type E, operation duration t, and historical similar accident rate p; According to the content of the work ticket, the risk feature vector F is constructed and the SVM classifier is used to classify the risk feature vector F through the radial basis kernel function. Calculate the similarity, compare the similarities under different risk levels based on the similarity, and output the work ticket risk level W. The formula is as follows: ; in, is the i-th standard risk feature vector under the j-th risk level in the risk level mapping library, which contains the feature information of the work ticket with known risk level. is the Lagrange multiplier, is the category label of the i-th sample in the training set, n is the number of samples in the training set, b is the bias term of the classification hyperplane, The final output work ticket risk level is determined by the sign function sign. Use support vector machine classifier to map the job risk level W∈{1-4}; Among them, level 1 low risk is inspection, level 2 general risk is ordinary equipment maintenance, level 3 higher risk is high-voltage equipment maintenance, and level 4 major risk is lifting operations.

[0009] In the preferred solution, the specific steps of step S3 are as follows: S31. Determination of equipment status data E: Based on the real-time three-dimensional coordinates (x, y, z) of step S20 and the static area electronic map of step S10, determine the static risk area level L and the equipment type set of the current area, and define the detection ROI; Get the video frame according to the detection ROI, preprocess the video frame, input the preprocessed video frame into the pre-trained YOLOv8 model, and output the device bounding box coordinates and category confidence ; According to the bounding box coordinates Crop the preprocessed video frame, input the device state classification head, and output the state probability , according to the state probability Get device status data E; The equipment status data E determines the running or shutdown status of the equipment through the probability threshold. The status determination formula is as follows: ; in, is the state determination threshold; The probability of device operation output by the model; S32. Determination of population density D: Collect video streams through cameras and count the number of people in the area based on the pre-trained YOLOv8 model. Calculate the real-time population density D and set the population threshold in the area is 1, if , then the calculation of the personnel density D is triggered, and the calculation formula is: ; Where N is the number of people and A is the area of the region.

[0010] In the preferred solution, the specific steps of step S3 are to fuse the static risk area data L, risk operation data W, equipment status data E and personnel density data D to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , using PSO algorithm to optimize weight parameters , the goal is to minimize the historical accident prediction error, the comprehensive risk value The calculation formula is: ; Among them, the weight parameter Through training of historical accident data, Reflect the nonlinear effect of population density; Based on the comprehensive risk value Determine risk level prediction results , as follows: ; Predict outcomes based on risk level Determine the final risk level , the specific formula is as follows: , take the static risk area level L and the risk level prediction result The highest value in .

[0011] In the preferred solution, the weight parameter The optimal weight is: 、 、 and .

[0012] In a preferred embodiment, the present invention further provides an intelligent warning device applied to the industrial safety risk classification warning method based on artificial intelligence dynamic priority as described in any of the above items, characterized in that it includes: a safety warning light sign body for mounting and protecting internal components; An AI processing module, configured to receive data from each module, calculate the risk level according to the algorithm in the industrial safety risk grading and warning method based on artificial intelligence dynamic priority according to any one of claims 1 to 4, and control other modules to perform corresponding warning actions; The LED light matrix consists of red, orange, yellow, and blue light strips, which are electrically connected to the AI processing module and present light warnings of different colors and flashing frequencies according to the risk level; The microphone is electrically connected to the AI processing module to transmit the collected sound signals to the AI processing module to assist in risk assessment; The SOS button is electrically connected to the AI processing module and is used to send a distress signal to the system; The speaker is electrically connected to the AI processing module and is used to make voice announcements based on risk levels and emergency situations; The clip is installed on one side of the safety warning light board body to fix the light board in a specific position; The charging socket is connected to the power management circuit inside the safety warning light body for charging the device; The vibration module is electrically connected to the AI processing module and is used to vibrate at specific frequencies and intensities at different risk levels; The GPS positioning module is electrically connected to the AI processing module to obtain the longitude and latitude information of the light sign location; The ultra-wideband positioning module is electrically connected to the AI processing module for indoor and outdoor positioning; The inertial measurement unit is electrically connected to the AI processing module to measure the acceleration and angular velocity information of the light sign, assist in positioning and abnormal situation detection; The 5G communication module is electrically connected to the AI processing module and is used to upload key data during emergency response; The acceleration sensor is electrically connected to the AI processing module to detect changes in the acceleration of the light sign and trigger the accident detection mechanism.

[0013] In a preferred embodiment, the present invention also provides a computer device / equipment / system, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of any one of the above-mentioned methods, devices, systems, storage media, and programs for industrial safety risk grading warnings based on artificial intelligence dynamic priorities.

[0014] In a preferred embodiment, the present invention further provides a computer non-transitory readable storage medium on which a computer program / instruction is stored, characterized in that when the computer program / instruction is executed by a processor, the steps of any of the above-mentioned industrial safety risk classification warning methods, devices, systems, storage media and programs based on artificial intelligence dynamic priority are implemented.

[0015] In a preferred embodiment, the present invention further provides a computer program product, including a computer program / instruction, characterized in that when the computer program / instruction is executed by one or more processors, it implements the steps of any of the above-mentioned industrial safety risk classification warning methods, devices, systems, storage media and programs based on artificial intelligence dynamic priority.

[0016] The present invention provides an industrial safety risk classification warning method, device, system, storage medium, and program based on artificial intelligence dynamic priority. Through the cooperation between the above structures, compared with the existing technology, it has the following beneficial effects: 1. By dividing the static risk areas into four levels based on historical accident data, equipment distribution, and 3D maps, the spatial risk baseline is clarified, providing accurate spatial benchmark data for dynamic risk assessment and enabling safety management to pre-judge risks in the spatial dimension. Second, it achieves real-time positioning of maintenance personnel and digitizes operational risks, establishes a database linking operational risks and personnel locations, and can obtain personnel locations and corresponding operational risk levels in real time. This provides real-time and accurate basic data support for dynamic risk assessment, solving the problem in static management where risk levels cannot be adjusted as personnel status changes. 3. Real-time collection of dynamic risk parameters such as equipment operating status and personnel density, combined with static risk area and operational risk data, forms a multi-dimensional risk assessment system. This effectively obtains real-time dynamic risk parameters, provides comprehensive and real-time data support for risk level prediction, and improves the timeliness and comprehensiveness of risk assessment. Fourth, the comprehensive risk value is calculated by integrating static risk, operational risk, and real-time dynamic parameters. The dynamic risk level is determined through a weighted scoring model and priority conflict resolution mechanism. This achieves a coupled assessment of multi-dimensional risk factors, solves the one-sidedness of single-factor assessment, and can more accurately reflect the actual risk situation on site. 5. Transform risk levels into visual, audible, and tactile multimodal warning signals. Through LED matrix color and flashing frequency, directional voice broadcast, and vibration modules, real-time risk notification is achieved, significantly improving personnel's risk perception efficiency and response speed, and enhancing the effectiveness of safety warnings, especially in complex industrial environments. 6. Through the dual accident detection mechanism of acceleration sensor threshold detection and SOS button triggering, combined with 5G data upload, linkage of surrounding warning lights, rapid equipment shutdown and full-process data tracing, rapid accident detection and emergency response are achieved, which greatly shortens the equipment shutdown response time and reduces accident losses. At the same time, it provides complete and reliable data records for post-analysis and safety management improvements, comprehensively improving the efficiency of emergency handling and the level of informationization of safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 This is the main structural diagram of the process of the present invention; Figure 2 It is a timing flow chart of the present invention; Figure 3 is a diagram of the safety management system of the present invention; Figure 4 It is a front view of the safety warning light sign device of the present invention; Figure 5 This invention Figure 4 Dorsal view of; Figure 6 is a schematic diagram of the structure of the computer device / equipment / system of the present invention; Figure 7 It is a structural schematic diagram of the static risk area of the present invention; Figure 8 It is a structural diagram of the system of the present invention.

[0018] Figure numerals: safety warning light sign body 1, LED light matrix 2, red light strip 21, orange light strip 22, yellow light strip 23, blue light strip 24, microphone 3, SOS button 4, speaker 5, clip 6, charging socket 7, vibration module 8, GPS positioning module 9, ultra-wideband positioning module 10, inertial measurement unit 11, 5G communication module 12, AI processing module 13, acceleration sensor 14. DETAILED DESCRIPTION

[0019] In order to better understand the purpose, structure and function of the present invention, the embodiments and features in the embodiments of the present invention can be combined with each other without conflict. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] Example 1 like Figures 1-3 As shown, an industrial safety risk classification warning method based on artificial intelligence dynamic priority includes the following steps: S10. Static risk area division: Obtain three-dimensional map data of the plant area, historical accident data, and equipment distribution area data, divide the plant area into four levels of static risk areas using the historical accident data and equipment distribution area data through a geo-fencing algorithm, determine the static risk area level L based on the four levels of static risk areas, and generate an electronic map of the static risk area containing coordinate boundaries based on the four levels of static risk areas.

[0021] Further, if Figure 2 and Figure 7 As shown, the specific steps of step S10 in this embodiment are as follows: S11. Data input: First, obtain high-precision three-dimensional map data of the factory area containing XYZ coordinate point clouds; Historical accident data set, where each historical accident data record the GPS coordinates (x, y, z) of the accident, the accident type T∈{fire, explosion, fall, electric shock}, and the severity S∈{1-4, 1 being the lightest}; Equipment distribution area data, including equipment type E∈{high-voltage equipment, lifting equipment, general machinery} and its coordinate range.

[0022] S12. Construction of accident hotspots and equipment risk fields: Geofencing algorithms with buffer zone analysis; The kernel density of historical accident coordinates is calculated based on the kernel density estimation (KDE) algorithm. The formula is: ; Where n is the number of accidents, h is the bandwidth (taken as 50m), d is the spatial dimension (3D), and K is the Gaussian kernel function.

[0023] The output generates the accident density heat map G(x,y,z).

[0024] Assign risk weights to equipment areas by type, i.e. high voltage equipment , lifting equipment , general machinery , generate equipment risk areas , the formula is as follows: ; in, Indicates the risk value of the equipment risk area at the accident density heat map G(x,y,z), reflecting the degree of safety risk caused by the presence of the equipment at that location. The risk weight of the i-th category equipment is a fixed value determined according to the type of equipment. For example, high-voltage equipment is more dangerous. The value is relatively large; the risk of ordinary machinery is relatively low. The smaller the value, the more it reflects the difference in inherent risks of different types of equipment. I is an indicator function used to determine whether the real-time three-dimensional coordinates (x, y, z) in the following step S20 are within the distribution area of a certain type of equipment in the accident density heat map G(x, y, z).

[0025] If the real-time 3D coordinates (x, y, z) are within the device area, I = 1, indicating that the location is affected by the device risk. If the real-time 3D coordinates (x, y, z) are not within the device area, I = 0, indicating that the location is not affected by the device risk. The value of I can be used to precisely define the scope of each device risk weight.

[0026] Regional clustering: Add the accident density heat map G(x, y, z) to the equipment risk area R(x, y, z), use the K-means clustering algorithm (set K=4) for spatial clustering, and combine it with the three-dimensional minimum bounding box algorithm to divide the static risk area into four levels. The static risk area level L∈{I-IV} is determined based on the four levels of static risk areas. The static risk area level L∈{I-IV} and the corresponding area types are as follows: Level I static risk areas are low-risk static areas, such as office areas and rest areas, with dense personnel but low equipment risk (L=1); Level II static risk areas are general risk static areas, belonging to ordinary operating areas and regular production activity areas (L=2); Level III static risk areas are high-risk static areas and belong to equipment maintenance areas, requiring close contact with equipment (L=3); Level IV static risk areas are high-risk static areas, belonging to hoisting / high-voltage areas, with immediate high-risk risks (L=4); S13. Output results Generate an electronic map of static risk areas based on the four-level static risk areas; The format of the static risk area electronic map is GeoJSON, including each area ID, coordinate boundary, and static risk area level L; In this embodiment, if Figure 1 As shown in the figure, after calculation, due to two hoisting wire rope breakage accidents in 2023 (coordinates (210,320,8) and (215,325,7)), superimposed on the distribution of high-voltage transformers, after kernel density and clustering calculation, this plant area is delineated as a Level IV risk area, and the boundary is marked as [(200,300,0),(250,350,15)].

[0027] S20. Positioning initialization and database establishment: Obtain the safety warning light device, use a multi-sensor fusion algorithm to output real-time three-dimensional coordinate data (x, y, z), obtain the work ticket content of the maintenance personnel, establish a work ticket risk level mapping library, and divide the work ticket content into four levels of risk operations: low-risk operations, general risk operations, large-risk operations, and major risk operations.

[0028] The specific steps are as follows: S21. Multi-sensor fusion positioning: Obtain a safety warning light sign device, integrate GPS, ultra-wideband (UWB) positioning module and inertial measurement unit (IMU) into the safety warning light sign device, fuse multi-source data through the extended Kalman filter algorithm (EKF), describe the position and speed changes of personnel movement through the state equation, and integrate the measurement values of each sensor through the observation equation. Finally, high-precision real-time three-dimensional coordinates (x, y, z) are output to ensure dynamic tracking of personnel position.

[0029] In this embodiment, the multi-sensor data of the safety warning light sign is input, including GPS (x, y, z), UWB positioning, and IMU (acceleration / angular velocity); Through the extended Kalman filter (EKF) fusion positioning, the calculation formula is as follows: Equation of state: ; Among them, x, y, z represent the position coordinates of GPS in three-dimensional space. are the velocities in the x, y, and z directions, respectively. ; is the acceleration in the x, y, and z directions, is the deterministic acceleration component.

[0030] Observation equation: ; Where H is the sensor observation matrix, is the process noise, is the observation noise; Output real-time 3D coordinates (x, y, z).

[0031] S22. Work Ticket Risk Mapping Obtain the maintenance personnel's work ticket content, including the operation type O∈{maintenance, installation, inspection}, the equipment involved E, the operation duration t, and the historical similar accident rate p; In this embodiment, a work ticket risk level mapping library is established, which stores a set of standard risk feature vectors corresponding to different risk levels. , construct the risk feature vector F according to the content of the work ticket, use the support vector machine (SVM) classifier, and use the radial basis kernel function Calculate the similarity. After calculating all similarities, compare the similarities under different risk levels. Classify the work ticket to be evaluated into the risk level with the greatest similarity. Output the work ticket risk level W. The formula is as follows: ; in, is the i-th standard risk feature vector under the j-th risk level in the risk level mapping library, which contains the feature information of the work ticket with known risk level. is the Lagrange multiplier, which is the parameter obtained by solving the optimization problem during SVM training. It is used to measure the importance of each training sample in the classification decision. is the category label of the i-th sample in the training set. If it belongs to a certain risk level, it is recorded as +1, and if it does not, it is recorded as -1. n is the number of samples in the training set. b is the bias term of the classification hyperplane, which is used to adjust the classification boundary position. The risk level of the work ticket is finally output, and its specific category is determined through the symbolic function.

[0032] Specifically, by mapping the standard feature vector of each risk level in the work ticket risk level library The weighted sum of similarities with the risk feature vector F to be classified, where the weight is , plus the bias b, and finally the risk level is determined by the sign function sign.

[0033] When the result is greater than 0, equal to 0, or less than 0, it corresponds to different risk level categories, that is, the work ticket risk level W∈{1-4}, realizing the digital conversion of operation risk; Among them, Level 1 low risk is inspection, Level 2 general risk is ordinary equipment maintenance, Level 3 high risk is high voltage equipment maintenance, and Level 4 major risk is lifting operation; In this embodiment, the digital conversion of operational risks is as follows: Inspection work → Level 1; General equipment maintenance → Level 2; High-voltage equipment maintenance → Level 3; Driving and hoisting operations → Level 4; In this embodiment, when a maintenance worker in this plant enters the Level III equipment maintenance area by carrying a safety warning light sign, the system confirms his coordinates as (190, 290, 4) through UWB positioning. At the same time, the work ticket is parsed as "transformer maintenance" and the rule engine determines it as Level 3 (high-risk operation), generating an initial risk parameter W=3.

[0034] S30, multi-source data collection: collecting equipment status data E and personnel density data D, and generating dynamic risk parameters based on the equipment status data E and personnel density data D; The specific steps are as follows: S31. Equipment status monitoring: Deploy industrial cameras in risk areas and use pre-trained YOLOv7 models to monitor equipment operating status in real time. S311, YOLOv8 model training (offline stage): Dataset collection: 100,000 equipment images were collected, including 60,000 images in operating state and 40,000 images in outage state, with equipment type (high voltage / hoisting / general machinery) and status label (operating / outage); Model improvement: Added a device status classification head to the YOLOv8 base model, including a fully connected layer and softmax, with an output dimension of 2 (operating / disabled); Training parameters: Adam optimizer, learning rate , batchsize=32, training 200 epochs, the verification set accuracy reached 98.2%.

[0035] S312, YOLOv8 model real-time detection process (online stage): ROI area positioning: Based on the real-time three-dimensional coordinates (x, y, z) of step S20, query the static area electronic map of step S10 to determine the static risk area level L and equipment type set of the current area, and delineate the detection ROI; Video frame preprocessing: Resize the input video frame to 640×640, normalize the pixel values to [0, 1], and add Gaussian noise to enhance robustness, where the noise standard deviation σ=0.05; YOLOv8 object detection: input the pre-processed video frame into the pre-trained improved YOLOv8 model and output the device bounding box coordinates and category confidence ; State classification: crop the image within the bounding box, input the device state classification head, and output the state probability ,like It is determined to be in running state Otherwise, it is in shutdown state , the state determination formula is as follows: ; in, is the state determination threshold; The device operation probability output by the model.

[0036] Furthermore, the real-time status of the device is determined by the probability threshold to ensure the reliability of the detection results and reduce the false positive rate.

[0037] In this example, in the Level III equipment maintenance area (coordinates (200, 300, 5)), the camera detects the rotation of a transformer cooling fan, and the YOLOv8 model outputs =0.92 (>0.8), confirming E=1, the equipment is in operation, with a confidence level of 92%.

[0038] S32. Calculation of population density: The population density is calculated by collecting video streams from cameras deployed in the area and using the YOLO model pre-trained in step S31 to detect the number of people in the area. , set the threshold of people in the area is 1, if , then trigger density calculation, real-time personnel density The calculation formula is as follows: ; Where N is the number of people in the area, counted by YOLO, is the area of the region, obtained from the static area electronic map in step S10; Output real-time personnel density with detection time and area ID .

[0039] Specifically, the degree of personnel gathering is quantified to provide parameters of personnel distribution dimensions for dynamic risk assessment, avoiding invalid calculations in single-person scenarios.

[0040] Example In the Level III equipment maintenance area, which is 50 m2 in area, the camera detects 5 people working at the same time, that is, N=5>1, and the calculation result is D=5 / 50=0.1 person / m2.

[0041] S40, dynamic risk level prediction: integrating static risk area data L, risk operation data W, equipment status data E and personnel density data D to generate a dynamic parameter set ; Among them, L is the static risk level (1-4), W is the operational risk level (1-4), E is the equipment status (0 / 1), and D is the personnel density (people / ㎡) Dynamically calculate the comprehensive risk value based on the dynamic parameter set , and use particle swarm optimization (PSO) to optimize the weight parameters , the goal is to minimize the historical accident prediction error, the comprehensive risk value The calculation formula is: ; Among them, the weight parameter Through training of historical accident data, Reflecting the nonlinear effect of population density, dense crowds may lead to chain accidents, for example, if =2, then its contribution value is 2 1.5 ≈2.828; The optimal weights for this embodiment are: 、 、 and , trained with 100,000 accident data, and based on the comprehensive risk value Determine risk level prediction results , as follows: ; Final grade determination: , take the highest level between static and predicted to avoid priority conflicts.

[0042] In this example, the level III equipment maintenance area has L=3, W=3 (higher risk operation), E=1 (equipment operation), and D=0.1. The calculation results show R=0.4×3+0.3×3+0.2×1+0.1×0.2=1.2+0.9+0.2+0.1=2.4. Since R<2.5, the predicted level is 3, so the final risk level is 3. S50, early warning and dynamic display: according to the final risk level Mapping to the color and flashing frequency corresponding to the LED matrix in the safety warning light device, where the LED matrix displays one of the four colors: red, orange, yellow, and blue; Based on the final risk level Acquire the voice and vibration modules in the safety warning light sign device, enhance personnel perception based on the voice and vibration modules, and synchronize them to the monitoring center via the 5G network; S51, LED matrix display rules are as follows:

[0043] S52, voice and vibration module: The voice module uses a 120dB directional speaker, according to the final risk level Trigger different contents. The trigger conditions are as follows: ; Vibration parameters: Final risk level At levels III to IV, a vibration of 0.2g to 0.8g is triggered for 3 seconds. When the LED matrix displays orange, the vibration is low-frequency, and when the LED matrix displays red, the vibration is high-frequency. The formula is as follows: , lasts for 3 seconds; In this embodiment, when the final level is Level III, the light board flashes orange at 2Hz, the voice broadcasts "Level III risk, wear protective equipment", and the vibration module triggers 0.2g low-frequency vibration.

[0044] S60, Emergency Response Linkage and Data Tracing: When an abnormal event is detected, the emergency event mechanism is triggered, as follows: S61, Accident Detection Mechanism: Triggering emergency events through dual detection, including acceleration sensor and SOS button; Acceleration sensor: detects acceleration exceeding 2g and lasting ≥5 seconds (triggering a fall / impact accident alarm); SOS button: A physical button on the back, which can be triggered by pressing and holding for 3 seconds to facilitate personnel to actively alarm; Output accident signal A∈{0,1}, A=1 indicates triggering.

[0045] S62, linkage operation 5G data upload: Real-time synchronization of workers' real-time 3D coordinate data (x, y, z), work ticket content, and static risk area L, ensuring instant synchronization of location and risk information, and uploading it to the security monitoring center, mobile phones, and tablets; Peripheral warning: The light sign within 50 meters of the accident point will flash red at 4Hz, and the voice broadcast will say "Emergency! Evacuate according to the evacuation route"; Equipment shutdown: The shutdown command is sent to the crane PLC via the Modbus protocol. The shutdown response time T is ≤ 300ms, which greatly improves personnel safety compared to the traditional manual operation of 30 seconds.

[0046] S63, Data Tracing Stored data: sensor raw data for the previous 10 minutes (including coordinate tracks, light board color logs, and device status curves) Report generation: Python script generates PDF files containing risk level heat maps, 3D coordinate annotations of accident points, and division of responsibility areas; In this example, an equipment leakage accident occurred in the Class III equipment maintenance area. The system was dual-triggered by the acceleration sensor and the maintenance personnel actively pressing the SOS button. The system shut down and uploaded data within 300ms. The traceability report showed that the equipment status E was continuously 1 (operating) for 5 minutes before the accident. Because the voice prompts the staff to wear protective equipment after entering the area, it was detected that the staff were wearing insulating gloves. This accident did not cause any casualties.

[0047] Example 2 The following describes a safety warning light sign device provided by the present invention and applied to the industrial safety risk grading warning method based on artificial intelligence dynamic priority described in Example 1. The safety warning light sign device based on the industrial safety risk grading warning method based on artificial intelligence dynamic priority described below and the industrial safety risk grading warning method based on artificial intelligence dynamic priority described above can be referenced to each other and further explained in conjunction with Example 1. like Figure 4 、 5 , the structure shown in 8, Figure 4 、 5 8 is a safety warning light sign device provided in an embodiment of the present application, which is an industrial safety risk grading warning method based on artificial intelligence dynamic priority. The safety warning light sign device in this embodiment is a key terminal device of the industrial safety risk grading warning system, and works closely with the industrial safety risk grading warning method based on artificial intelligence dynamic priority to realize real-time positioning of personnel in industrial scenes, risk warnings, and emergency response functions to ensure personnel safety, including: a safety warning light sign body 1, an LED light matrix 2, a microphone 3, an SOS button 4, a speaker 5, a clip 6, a charging socket 7, a vibration module 8, a GPS positioning module 9, an ultra-wideband positioning module 10, an inertial measurement unit 11, a 5G communication module 12, an AI processing module 13, and an acceleration sensor 14. Each module works together to provide all-round support for industrial safety.

[0048] The safety warning light body 1 serves as the physical carrier of the entire device, providing a mounting base for other components and protecting the internal circuits and modules. Its housing is made of high-strength, fire-resistant, and corrosion-resistant materials, making it suitable for harsh industrial environments.

[0049] The LED light matrix 2 is composed of a red light strip 21, an orange light strip 22, a yellow light strip 23, and a blue light strip 24. The red light strip 21, the orange light strip 22, the yellow light strip 23, and the blue light strip 24 are arranged on one side of the safety warning light sign body 1, and the red light strip 21, the orange light strip 22, the yellow light strip 23, and the blue light strip 24 are arranged at equal intervals.

[0050] The LED lamp matrix 2 is electrically connected to the AI processing module 13 and receives control signals from the AI processing module 13 .

[0051] Specifically, the LED light matrix 2 presents light warnings of different colors and flashing frequencies according to the risk level. For example, at level I risk, the blue light strip 24 is constantly on at 1Hz; at level II risk, the yellow light strip 23 is constantly on at 1Hz; at level III risk, the orange light strip 22 flashes at 2Hz; at level IV risk, the red light strip 21 flashes at 2Hz, providing on-site personnel with intuitive visual risk prompts.

[0052] The microphone 3 is electrically connected to the AI processing module 13 and transmits the collected sound signal to the AI processing module 13 . The microphone 3 is arranged on one side of the safety warning light sign body 1 .

[0053] Specifically, microphone 3 is used to collect environmental sounds, which can assist in determining whether there are abnormal sounds at the scene, such as abnormal noise from equipment, people shouting for help, etc., providing additional data support for risk assessment.

[0054] The SOS button 4 is electrically connected to the AI processing module 13 , and sends an emergency signal to the AI processing module 13 when pressed. The SOS button 4 is provided on one side of the safety warning light board body 1 .

[0055] Specifically, when people encounter an emergency or dangerous situation, a long press for 3 seconds can trigger an emergency response, send a distress signal to the system, and initiate relevant emergency measures, such as accident detection mechanisms and linkage operations.

[0056] The speaker 5 is electrically connected to the AI processing module 13 and receives the voice commands sent by the AI processing module 13 .

[0057] Specifically, voice broadcasts are made according to the risk level and emergency situation. For example, in case of Level III risk, the broadcast is "Level III risk, wear protective equipment"; in case of Level IV risk, the broadcast is "Level IV risk, evacuate immediately!"; in case of emergency response, the broadcast is "Emergency! Evacuate according to the evacuation route" and other prompt information to give personnel clear voice guidance.

[0058] The clip 6 is installed on the side of the safety warning light board body 1 away from the LED light matrix 2, and is used to fix the light board on the personnel's clothing or equipment to ensure that the light board moves with the personnel to achieve real-time positioning and warning functions.

[0059] The charging socket 7 is connected to the power management circuit inside the safety warning light sign body 1 and is used to charge the device to ensure continuous power supply to the device and maintain normal operation.

[0060] The vibration module 8 is electrically connected to the AI processing module 13 and receives the vibration control signal from the AI processing module. The vibration module 8 is disposed inside the safety warning light sign body 1 .

[0061] Specifically, at level III risk, the vibration module 8 vibrates at a low frequency of 0.2g (5Hz); at level IV risk, it vibrates at a high frequency of 0.8g (20Hz). Vibration is used to remind people of risk situations, which is especially suitable for scenarios in noisy environments where people cannot hear voice warnings or see light warnings in time.

[0062] The GPS positioning module 9 is electrically connected to the AI processing module 13 and transmits positioning data to the AI processing module 13 .

[0063] Specifically, it provides global positioning services to obtain the latitude and longitude information of the location of the light sign for personnel positioning, with an accuracy that can reach the civilian GPS positioning accuracy range. After being integrated with the data of the ultra-wideband positioning module 10 and the inertial measurement unit 11, the positioning accuracy can be further improved.

[0064] The ultra-wideband (UWB) positioning module 10 is electrically connected to the AI processing module 13 and sends high-precision positioning data to the AI processing module 13 .

[0065] Specifically, ultra-wideband technology is used to achieve high-precision indoor and outdoor positioning, with a positioning accuracy of ±0.1m. Working in conjunction with the GPS positioning module 9 and the inertial measurement unit 11, it provides precise location data for personnel positioning initialization, meeting the high-precision positioning requirements of industrial scenarios.

[0066] The inertial measurement unit 11 is electrically connected to the AI processing module 13 and transmits the measured acceleration and angular velocity data to the AI processing module 13 .

[0067] Specifically, the acceleration and angular velocity information of the light sign is measured in real time. Through the data fusion of the extended Kalman filter (EKF) algorithm with the GPS positioning module 9 and the ultra-wideband (UWB) positioning module 10, more accurate real-time three-dimensional coordinate (x, y, z) positioning of personnel can be achieved, with an accuracy of up to ±0.3m. At the same time, it can assist in detecting the movement status of personnel and whether abnormal situations such as falling and collision have occurred.

[0068] The 5G communication module 12 is electrically connected to the AI processing module 13 for high-speed data transmission.

[0069] Specifically, during emergency response, key data such as real-time coordinates (x, y, z), work ticket content, static risk area L, etc. are uploaded to the background system with a transmission delay of ≤100ms, ensuring fast and stable data transmission and providing timely data support for emergency decision-making.

[0070] The AI processing module 13 is the core processing unit of the device and is electrically connected to all other modules. The AI processing module 13 receives positioning data from the GPS positioning module (9), the ultra-wideband (UWB) positioning module 10, the inertial measurement unit 11, the sound data from the microphone 3, the acceleration data from the acceleration sensor 14, and the operation risk data transmitted from the external system such as the work ticket; Based on these data, combined with the algorithm in the industrial safety risk grading warning method based on artificial intelligence dynamic priority in Example 1, the risk level is calculated, and the LED light matrix 2, speaker 5, vibration module 8, etc. are controlled to perform corresponding warning actions.

[0071] The acceleration sensor 14 is electrically connected to the AI processing module 13 and transmits the measured acceleration data to the AI processing module.

[0072] Specifically, the acceleration changes of the light sign are detected. If the acceleration value exceeds 2g and lasts for ≥5 seconds (triggering a fall / impact accident alarm), the accident detection mechanism is triggered, an accident signal is sent to the AI processing module, and the emergency response linkage operation is initiated.

[0073] Through the close collaboration of the above components, this safety warning light device can perceive environmental information in real time, accurately determine the risk level of the personnel's location, and issue warnings to personnel in various ways. At the same time, it can respond quickly in emergency situations, providing strong protection for industrial production safety.

[0074] Example 3 Further illustrate with reference to Examples 1 and 2. Figure 6 The structure shown, Figure 6 This is a schematic diagram of the structure of a computer device / equipment / system provided in an embodiment of the present application. The computer device / equipment / system includes: A processor, memory, a communication bus, and a computer program stored in the memory and executable on the processor.

[0075] The processor can call the computer program in the memory, and when executing the program, implement an industrial safety risk grading warning method, device, system, storage medium and program based on artificial intelligence dynamic priority provided in the above embodiment, the method including: S10, static risk area division: determine the static risk area level L based on the three-dimensional map data of the factory area, historical accident data and equipment distribution area data; S20, positioning initialization and database establishment: output real-time three-dimensional coordinate data (x, y, z) according to the multi-sensor fusion algorithm, establish a work ticket risk level mapping library, and map the operation risk level W in combination with the work ticket content; S30, multi-source data acquisition: generate dynamic risk parameters based on the pre-trained YOLOv8 model; S40, dynamic risk level prediction: integrate the static risk area level L, the operation risk level W and the dynamic risk parameters to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , predicting dynamic risk level ;According to the dynamic risk level Determine the final risk level ; S50, early warning and dynamic display: according to the final risk level Trigger multimodal warning signals, including LED display, voice prompts and vibration feedback; S60, emergency response linkage and data tracing: When an abnormal event is detected, the emergency event mechanism is triggered.

[0076] Furthermore, the computer device / equipment / system also includes: Communications Interface: used for communication between memory and processor.

[0077] The memory may include a high-speed RAM memory and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0078] If the memory, processor, and communication interface are implemented independently, the communication interface, memory, and processor can be interconnected via a bus to facilitate communication. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses. For ease of illustration, the diagram uses a single thick line, but this does not imply a single bus or type of bus.

[0079] In addition, the logical instructions in the above-mentioned memory can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0080] A processor may include one or more processing units. For example, a processor may include an application processor (AP), an application-specific integrated circuit (ASIC), a modem processor, a central processing unit (CPU), an image signal processor (ISP), a controller, memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU). The different processing units may be independent devices or integrated into one or more processors. The controller may be a neural network center or command center. The controller may generate operation control signals based on instruction opcodes and timing signals to control the retrieval and execution of instructions. The processor may also include memory for storing instructions and data. In some embodiments, the memory in the processor is a high-speed cache memory. This memory can store instructions or data that have just been used or are being recycled by the processor. If the processor needs to use the instruction or data again, it can directly call it from the memory. This avoids repeated accesses, reduces processor latency, and thus improves system efficiency.

[0081] A visualization module is used to display images, videos, etc. The visualization module may include a display panel, which may be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode or an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), MiniLED, MicroLed, Micro-oLed, or a quantum dot light-emitting diode (QLED).

[0082] Optionally, in a specific implementation, if the memory, processor, and communication interface are integrated on a chip, the memory, processor, and communication interface can communicate with each other through an internal interface.

[0083] On the other hand, an embodiment of the present application also provides a computer non-transitory readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned industrial safety risk grading warning method, device, system, storage medium and program based on dynamic priority of artificial intelligence, the method comprising: S10, static risk area division: determining the static risk area level L based on the three-dimensional map data of the factory area, historical accident data and equipment distribution area data; S20, positioning initialization and database establishment: outputting real-time three-dimensional coordinate data (x, y, z) according to the multi-sensor fusion algorithm, establishing a work ticket risk level mapping library, and mapping the operation risk level W in combination with the work ticket content; S30, multi-source data acquisition: generating dynamic risk parameters based on the pre-trained YOLOv8 model; S40, dynamic risk level prediction: integrating the static risk area level L, the operation risk level W and the dynamic risk parameters to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , predicting dynamic risk level ;According to the dynamic risk level Determine the final risk level ; S50, early warning and dynamic display: according to the final risk level Trigger multimodal warning signals, including LED display, voice prompts and vibration feedback; S60, emergency response linkage and data tracing: When an abnormal event is detected, the emergency event mechanism is triggered.

[0084] On the other hand, an embodiment of the present application also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. The computer program can run computer instructions. When the computer program is executed by a processor, the computer can execute an industrial safety risk grading warning method, device, system, storage medium and program based on artificial intelligence dynamic priority provided by the above methods. The method includes: S10, static risk area division: determining the static risk area level L based on the three-dimensional map data of the factory area, historical accident data and equipment distribution area data; S20, positioning initialization and database establishment: outputting real-time three-dimensional coordinate data (x, y, z) according to the multi-sensor fusion algorithm, establishing a work ticket risk level mapping library, and mapping the operation risk level W in combination with the work ticket content; S30, multi-source data acquisition: generating dynamic risk parameters based on the pre-trained YOLOv8 model; S40, dynamic risk level prediction: integrating the static risk area level L, the operation risk level W and the dynamic risk parameters to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , predicting dynamic risk level ;According to the dynamic risk level Determine the final risk level ; S50, early warning and dynamic display: according to the final risk level Trigger multimodal warning signals, including LED display, voice prompts and vibration feedback; S60, emergency response linkage and data tracing: When an abnormal event is detected, the emergency event mechanism is triggered.

[0085] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program can be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0086] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. An industrial safety risk classification warning method based on artificial intelligence dynamic priority, characterized in that: The following steps are involved: S10. Static risk area classification: Determine the static risk area level L based on the plant's three-dimensional map data, historical accident data, and equipment distribution area data; S20, positioning initialization and database establishment: Output real-time three-dimensional coordinate data (x, y, z) based on the multi-sensor fusion algorithm, establish a work ticket risk level mapping library, and map the operation risk level W based on the work ticket content; S30, Multi-source data collection: Generate dynamic risk parameters based on the pre-trained YOLOv8 model; S40, dynamic risk level prediction: integrating the static risk area level L, the operation risk level W and the dynamic risk parameters to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , predicting dynamic risk level ;According to the dynamic risk level Determine the final risk level ; S50, early warning and dynamic display: according to the final risk level Trigger multi-modal warning signals, including LED display, voice prompts and vibration feedback; S60, Emergency response linkage and data tracing: When an abnormal event is detected, the emergency event mechanism is triggered.

2. The industrial safety risk classification warning method based on artificial intelligence dynamic priority according to claim 1 is characterized in that: The specific steps of step S10 are as follows: S11. Input the factory 3D map data, historical accident GPS coordinates and equipment distribution area data; S12. Construct an accident density heat map G(x,y,z) using the KDE algorithm and assign risk weights to each device type to generate device risk areas. , the formula is as follows: ； Where I is the indicator function, To represent the risk weight of the i-th category equipment, it is a fixed value determined according to the equipment type; S13. Overlay the accident density heat map G(x, y, z) with the equipment risk area R(x, y, z), and use the K-means clustering algorithm combined with the three-dimensional minimum bounding box algorithm to divide the four-level static risk areas, generating a static area electronic map in GeoJSON format containing coordinate boundaries; Among them, the static risk level L∈{I-IV} is determined according to the four-level static risk area. The level I static risk area is a low-risk static area, the level II static risk area is a general risk static area, the level III static risk area is a large risk static area, and the level IV is a high-risk static area.

3. The industrial safety risk classification warning method based on artificial intelligence dynamic priority according to claim 1 is characterized in that: The specific steps of step S2 are as follows: S21. Multi-sensor fusion positioning: Obtain a safety warning light device that integrates GPS, UWB, and IMU sensors, and output real-time three-dimensional coordinates (x, y, z) through the EKF algorithm. The calculation formula is as follows: Equation of state: ； Among them, x, y, z represent the position coordinates of GPS in three-dimensional space. are the velocities in the x, y, and z directions, respectively. ; is the acceleration in the x, y, and z directions, is the deterministic acceleration component; Observation equation: ; Where H is the sensor observation matrix, is the process noise, is the observation noise; S22. Work Ticket Risk Mapping: Establish a work ticket risk level mapping library, and obtain a set of standard risk feature vectors based on the work ticket risk level mapping library. ; Obtain work ticket content, including operation type O, equipment type E, operation duration t, and historical similar accident rate p; According to the content of the work ticket, the risk feature vector F is constructed and the SVM classifier is used to classify the risk feature vector F through the radial basis kernel function. Calculate the similarity, compare the similarities under different risk levels based on the similarity, and output the work ticket risk level W. The formula is as follows: ； in, is the i-th standard risk feature vector under the j-th risk level in the risk level mapping library, which contains the feature information of the work ticket with known risk level. is the Lagrange multiplier, is the category label of the i-th sample in the training set, n is the number of samples in the training set, b is the bias term of the classification hyperplane, The final output work ticket risk level is determined by the sign function sign. Use support vector machine classifier to map the job risk level W∈{1-4}; Among them, level 1 low risk is inspection, level 2 general risk is ordinary equipment maintenance, level 3 higher risk is high-voltage equipment maintenance, and level 4 major risk is lifting operations.

4. The industrial safety risk classification warning method based on artificial intelligence dynamic priority according to claim 2 is characterized in that: The specific steps of step S3 are as follows: S31. Determination of equipment status data E: Based on the real-time three-dimensional coordinates (x, y, z) of step S20 and the static area electronic map of step S10, determine the static risk area level L and the equipment type set of the current area, and define the detection ROI; Get the video frame according to the detection ROI, preprocess the video frame, input the preprocessed video frame into the pre-trained YOLOv8 model, and output the device bounding box coordinates and category confidence ; According to the bounding box coordinates Crop the preprocessed video frame, input the device state classification head, and output the state probability , according to the state probability Get device status data E; The equipment status data E determines the running or shutdown status of the equipment through the probability threshold. The status determination formula is as follows: ； in, is the state determination threshold; The probability of device operation output by the model; S32. Determination of population density D: Collect video streams through cameras and count the number of people in the area based on the pre-trained YOLOv8 model. Calculate the real-time population density D and set the population threshold in the area is 1, if , then the calculation of the personnel density D is triggered, and the calculation formula is: ； Where N is the number of people and A is the area of the region.

5. The industrial safety risk classification warning method based on artificial intelligence dynamic priority according to claim 1 is characterized in that: The specific steps of step S3 are to fuse the static risk area data L, risk operation data W, equipment status data E and personnel density data D to generate a dynamic parameter set ; According to the dynamic parameter set Dynamic calculation of comprehensive risk value , using PSO algorithm to optimize weight parameters , the goal is to minimize the historical accident prediction error, the comprehensive risk value The calculation formula is: ； Among them, the weight parameter Through training of historical accident data, Reflect the nonlinear effect of population density; Based on the comprehensive risk value Determine the risk level prediction results , as follows: ； Predict outcomes based on risk level Determine the final risk level , the specific formula is as follows: , take the static risk area level L and the risk level prediction result The highest value in .

6. The industrial safety risk classification warning method based on artificial intelligence dynamic priority according to claim 5 is characterized in that: Weight parameters The optimal weight is: 、 、 and .

7. An intelligent warning device applied to the industrial safety risk classification warning method based on artificial intelligence dynamic priority as described in any one of claims 1 to 6, characterized in that: include: A safety warning light board body (1) is used to install and protect internal components; An AI processing module (13) is used to receive data from each module, calculate the risk level according to the algorithm in the industrial safety risk classification warning method based on artificial intelligence dynamic priority according to any one of claims 1 to 6, and control other modules to perform corresponding warning actions; The LED light matrix (2) is composed of a red light strip (21), an orange light strip (22), a yellow light strip (23), and a blue light strip (24), and is electrically connected to the AI processing module (13) to present light warnings of different colors and flashing frequencies according to the risk level; The microphone (3) is electrically connected to the AI processing module (13) and is used to transmit the collected sound signals to the AI processing module to assist in risk assessment; The SOS button (4) is electrically connected to the AI processing module (13) and is used to send a distress signal to the system; The loudspeaker (5) is electrically connected to the AI processing module (13) and is used to make voice announcements according to the risk level and emergency situation; A clip (6) is mounted on one side of the safety warning light board body (1) and is used to fix the light board at a specific position; The charging socket (7) is connected to the internal power management circuit of the safety warning light board body (1) for charging the device; The vibration module (8) is electrically connected to the AI processing module (13) and is used to vibrate at a specific frequency and intensity under different risk levels; The GPS positioning module (9) is electrically connected to the AI processing module (13) to obtain the longitude and latitude information of the light sign position; The ultra-wideband positioning module (10) is electrically connected to the AI processing module (13) for indoor and outdoor positioning; The inertial measurement unit (11) is electrically connected to the AI processing module (13) and is used to measure the acceleration and angular velocity information of the light sign, and assist in positioning and abnormality detection; The 5G communication module (12) is electrically connected to the AI processing module (13) for uploading key data during emergency response; The acceleration sensor (14) is electrically connected to the AI processing module (13) and is used to detect changes in the acceleration of the light sign and trigger an accident detection mechanism.

8. A computer device / apparatus / system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the computer program to implement the steps of the industrial safety risk classification warning method based on artificial intelligence dynamic priority as described in any one of claims 1-6.

9. A computer non-transitory readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the industrial safety risk classification warning method based on artificial intelligence dynamic priority as described in any one of claims 1-6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by one or more processors, the steps of the industrial safety risk classification warning method based on artificial intelligence dynamic priority as described in any one of claims 1 to 6 are implemented.

Images