Chemical industrial park safety risk intelligent management and control system and method

Abstract

The application discloses a chemical industrial park safety risk intelligent management and control system and method, belongs to the technical field of risk management and control, and solves the problems of incomplete tank area risk perception, data transmission delay and insufficient dynamic path risk avoidance. The system comprises a sensing device, a 5G communication module and an industrial Ethernet, a data transmission network is formed by the 5G communication module and the industrial Ethernet, and data is transmitted to a cloud platform; a time sequence database is used by the cloud platform to establish a device code index, a rule engine executes leakage judgment, and when methane is continuously 3-5 cycles > 25% LEL and the pressure drop rate > 0.5 MPa / min, a warning is triggered; a risk analysis module calls three-dimensional geographic information data to generate a 500-meter-2000-meter radius evacuation path through a Dijkstra algorithm, avoids the upwind area of the real-time wind direction and preferentially connects a fire station; and an application terminal receives a warning instruction containing a leakage device code and a dynamic path. The system is used for rapid warning and evacuation planning of chemical safety accidents.

Description

Technical Field

[0001] This invention relates to the field of risk management technology for chemical industrial parks, specifically to an intelligent management system and method for safety risks in chemical industrial parks. Background Technology

[0002] Safety risk monitoring in tank areas of major hazardous sources in chemical industrial parks suffers from delayed response and low evacuation efficiency, primarily due to a combination of technological deficiencies. At the risk perception level, existing systems typically employ only a single type of sensor for monitoring, such as relying solely on gas concentration or pressure detection, making it difficult to capture risks coupled with multiple parameters. Since early gas leaks often exhibit a combination of sudden pressure drops and gradual concentration increases, a single sensor cannot establish correlation rules between parameters, resulting in a persistently high false negative rate for early leak warnings. Furthermore, traditional sensors are often fixed externally to the tank, leading to slow responses to localized concentration gradient changes caused by minor leaks, delaying optimal response times.

[0003] In the data transmission stage, existing systems generally rely on a single network standard for communication. When using wireless networks, the electromagnetic shielding effect generated by the dense metal tanks in chemical industrial parks easily causes signal attenuation, resulting in frequent data packet loss during high-concurrency data transmission, making it difficult to fully upload the changing trends of key parameters (such as the instantaneous rate of pressure drop). On the other hand, when using wired industrial buses, network scalability is poor, new monitoring points require rewiring, and bandwidth allocation is rigid, unable to adapt to traffic peaks under sudden risk events. These shortcomings lead to timestamp misalignment or missing key frames in the data received by the cloud platform, directly affecting the timeliness of subsequent risk analysis.

[0004] During the risk management phase, existing route planning methods have significant limitations. Firstly, evacuation routes rely on static electronic map data and lack real-time integration of meteorological monitoring information. Since the spread of chemical gases is directly affected by wind direction and speed, static routes may guide personnel into the gas diffusion area, causing secondary injuries. Secondly, route planning is not dynamically linked to emergency resources; for example, it does not prioritize connections to the nearest available fire station or medical point, resulting in detours for rescue personnel reaching the leak point and delaying initial response efficiency. Furthermore, evacuation instructions issued by traditional systems lack detailed spatial information, providing only text warnings or simple diagrams, making it impossible for on-site personnel to quickly determine their location and the geometric relationship to the optimal route.

[0005] The aforementioned problems stem from three technical bottlenecks: First, real-time fusion of multi-source heterogeneous sensor data requires solving the parallel processing challenge of high-throughput time-series data, especially since parameters such as gas concentration, pressure, and temperature have different sampling frequencies and dimensions, and establishing cross-parameter association rules requires overcoming the complexity of time series alignment and feature extraction. Second, dynamic path planning needs to overcome the computational barrier of efficient coupling between meteorological data and geographic information systems; the impact of wind direction changes on paths requires a second-level response, but traditional spatial analysis algorithms have high computational overhead. Finally, the synchronous issuance of early warning instructions and evacuation routes requires balancing transmission reliability and real-time performance; high-precision vector maps have large data volumes, and achieving low-latency transmission under limited bandwidth presents a technical contradiction. These difficulties have long constrained the improvement of safety management efficiency in chemical industrial parks. Summary of the Invention

[0006] This invention provides an intelligent safety risk management system and method for chemical industrial parks, aiming to solve the problems of incomplete multi-parameter risk perception, insufficient real-time data transmission, and lack of integration of real-time meteorological and emergency resources into dynamic evacuation routes in tank areas of major hazardous sources in chemical industrial parks. Existing systems suffer from high false alarm rates for leak warnings, evacuation routes traversing hazardous areas, and delayed rescue responses due to single sensors, rigid network architecture, and static route planning. This invention addresses the problem of evacuation routes failing due to obstacles. Traditional systems rely on preset maps; adding obstacles (such as collapsed facilities) triggers poor timeliness in route replanning, failing to guarantee route availability. It also addresses the problem of congestion in issuing warning commands when multiple terminals access the system concurrently. Existing broadcast distribution does not classify risks according to proximity; terminals in high-risk areas may experience delays in receiving critical route updates due to network congestion. Furthermore, it addresses the problem of slow response times in manual handling processes. Traditional methods rely on manual judgment of leak characteristics and manual route drawing, taking more than minutes from data acquisition to command issuance, failing to meet the requirements of second-level emergency response. Finally, it addresses the problem of dynamic weather changes weakening route safety. When wind speed and direction change, fixed-angle avoidance areas cannot match the gas diffusion range, causing the route to still pass through contaminated areas. Addressing the issue of low efficiency in coordinated evacuation from multiple leak points: Independent route planning fails to consider the aggregation effect of leak points, potentially leading to overlapping and conflicting routes, increasing overall evacuation time. Addressing the risk of increased congestion on main evacuation routes: Lack of automatic diversion mechanisms when personnel density exceeds limits; traditional systems only provide a single path, easily creating bottlenecks. Addressing the issue of personnel deviation in on-site route execution: Personnel are prone to disorientation in complex environments; static route maps cannot dynamically correct movement trajectories, and there is no real-time intervention after deviation. Addressing the issue of heavy protective equipment hindering personnel's perception of alarms: Soundproofing chemical protective suits render voice prompts ineffective; vibration feedback is not adapted to equipment weight differences, potentially leading to missed critical turning commands. Addressing the issue of decreased alarm signal recognition rate during vigorous movement: Running and bumping reduce the perceptibility of conventional vibration / light signals, and there is no dynamic compensation for motion interference.

[0007] To achieve these objectives and other advantages according to the present invention, an intelligent management and control system for safety risks in chemical industrial parks is provided, including sensing layer devices, data transmission networks, cloud platform data processing centers, and application terminals deployed within the chemical industrial park;

[0008] The sensing layer equipment includes gas concentration sensors, temperature sensors, pressure sensors, and flame detectors installed in the tank area of ​​major hazardous sources;

[0009] The data transmission network adopts a dual-mode transmission architecture of 5G communication module and industrial Ethernet. The sensing layer device packages the collected data and transmits it to the cloud platform data processing center.

[0010] The cloud platform data processing center includes a time-series database, a rule engine, and a risk analysis module. The time-series database stores the collected data from the sensing layer devices in timestamp format and establishes a device code index. The rule engine has preset leakage judgment rules: when the methane concentration data associated with the same device code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data decrease rate is greater than 0.5 MPa / min, a leakage warning is triggered. After receiving the warning signal output by the rule engine, the risk analysis module calls the device coordinate data, real-time meteorological data, and emergency resource distribution data from the 3D geographic information system, and calculates the evacuation path within a radius of 500-2000 meters centered on the leak point using the Dijkstra algorithm. The evacuation path avoids the upwind area of ​​the real-time wind direction and prioritizes connecting the coordinates of the fire station.

[0011] The application terminal receives early warning instructions and evacuation route vector maps issued by the cloud platform data processing center. The early warning instructions include the leakage device code, early warning level, and gas diffusion simulation polygons. The evacuation route vector maps include a dynamically updated set of path node coordinates and the estimated travel time for each node.

[0012] Preferably, in the intelligent management and control system for safety risks in chemical industrial parks of the present invention, the risk analysis module further includes a lidar scanning unit, which continuously acquires real-time point cloud data of the path node area after generating the evacuation path; the lidar scanning unit has a scanning frequency of 5Hz-10Hz and a scanning angle range covering 30-60 degrees on both sides of the center line of the evacuation path.

[0013] The risk analysis module performs a differential comparison between real-time point cloud data and the base map elevation data of the 3D geographic information system. When the absolute difference between the surface point cloud elevation value of a newly added obstacle and the base map elevation data is greater than 50 centimeters, the evacuation path replanning is triggered. The replanning process uses the A-Star algorithm to recalculate the obstacle avoidance path, and the updated evacuation path vector map is synchronized to the application terminal within 3 to 8 seconds.

[0014] Preferably, in the intelligent management and control system for safety risks in chemical industrial parks of the present invention, the cloud platform data processing center further includes an instruction distribution module connected to both the risk analysis module and the application terminal. The instruction distribution module includes:

[0015] The terminal positioning unit obtains physical location data through the GPS positioning module built into the application terminal, with a positioning accuracy error of less than 3 meters;

[0016] The distance classification unit calculates the straight-line distance between each application terminal and the leak point, and divides it into three priority levels. The first priority is a distance of less than or equal to 500 meters from the leak point, the second priority is a distance of more than 500 meters but less than or equal to 1000 meters from the leak point, and the third priority is a distance of more than 1000 meters but less than or equal to 2000 meters from the leak point. The warning is automatically lifted when the distance is greater than 2000 meters.

[0017] When evacuation route replanning is triggered, the transmission control unit directly transmits the evacuation route vector map to the first priority application terminal via UDP protocol through a 5G communication module. Data packets transmitted to the second and third priority application terminals are stored in a Redis cache queue and then transmitted via industrial Ethernet after a delay of 2-5 seconds. The resource allocation ratio of the transmission control unit is dynamically adjusted according to the personnel evacuation priority strategy in the emergency plan of the chemical industrial park. The first priority application terminal is allocated 60%-80% of the total bandwidth, and the remaining bandwidth is allocated according to the ratio of the number of second priority application terminals to the number of third priority application terminals = 3:1.

[0018] This invention also provides an intelligent management and control method for safety risks in chemical industrial parks, comprising the following steps:

[0019] Step 1: Collect methane concentration data in real time using gas concentration sensors deployed in the tank area of ​​major hazardous sources;

[0020] Step 2: Collect temperature data of the tank area using temperature sensors;

[0021] Step 3: Collect tank area pressure data using pressure sensors;

[0022] Step four: Collect flame radiation data using a flame detector;

[0023] Step 5: The methane concentration data, temperature data, pressure data, and flame radiation data collected in steps 1 to 4 are sent to the cloud platform data processing center via a 5G communication module and an industrial Ethernet dual-mode transmission architecture.

[0024] Step 6: Store the received data in the time-series database of the cloud platform data processing center, and establish a device code index in timestamp format;

[0025] Step 7: Perform leak detection through the rule engine: When the methane concentration data associated with the same equipment code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data drop rate is greater than 0.5 MPa / min, a leak warning signal is generated.

[0026] Step 8: Respond to the leak warning signal and call up the equipment coordinate data, real-time meteorological data, and emergency resource distribution data stored in the 3D geographic information system;

[0027] Step 9: Based on real-time wind direction data, set the evacuation route avoidance area. The evacuation route area covers a fan-shaped area 30-60 degrees upwind of the leak point and is 200-800 meters away from the leak point.

[0028] Step 10: Calculate evacuation routes within a radius of 500-2000 meters centered on the leak point using the Dijkstra algorithm. The evacuation routes will bypass the evacuation route avoidance area set in Step 9 and will preferentially connect to the coordinates of the nearest fire station.

[0029] Step 11: Encapsulate the leaking device code, warning level, and gas diffusion simulation polygon into a warning command;

[0030] Step 12: Encapsulate the coordinate set of evacuation route nodes, the estimated travel time of each node, and the real-time wind direction vector into an evacuation route vector map.

[0031] Step 13: Send early warning commands and evacuation route vector maps to application terminals via the 5G communication module.

[0032] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, step eight further includes: extracting wind speed data and wind direction angle data from real-time meteorological data, wherein the wind speed detection range is 1 m / s to 30 m / s and the wind direction angle detection range is 0 degrees to 360 degrees; performing cubic spline interpolation calculation on the wind speed data and wind direction angle data with an interpolation period of 15 seconds to 40 seconds to generate a continuous time series wind field model; and associating the output of the wind field model with the leakage point equipment code and storing it as dynamic meteorological vector data.

[0033] The path avoidance area set in step nine is updated based on dynamic meteorological vector data. When the wind speed data is 8 m / s-15 m / s, the angle of the fan-shaped area expands to the range of 45 degrees-90 degrees.

[0034] In step ten, the calculation frequency of evacuation routes is increased to once every 10-30 seconds.

[0035] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, step eight further includes: retrieving the equipment codes associated with all current leak warning signals, extracting the coordinate data of each leak point and dynamic meteorological vector data; establishing a buffer zone with a radius of 30-60 meters centered on the leak point coordinates, and using a spatial overlay analysis algorithm to detect the overlapping part of the buffer zone; when the overlapping area accounts for more than 40%-70%, merging the dynamic meteorological vector data and generating a joint path avoidance area;

[0036] In step nine, the range of the path avoidance area is expanded to: a 30-60 degree sector area for a single leak point; and a 50-120 degree composite sector area for a combined path avoidance area, with the distance taken as the maximum range of 200-800 meters among all leak points.

[0037] In step ten, the evacuation route calculation adopts a hierarchical planning strategy: first, the main evacuation route is planned, connecting the coordinates of the safety exits within a radius of 500-2000 meters; then, branch routes are generated based on the main evacuation route to the coding areas of each equipment.

[0038] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, the branch path generation stage in step ten includes the following steps:

[0039] S1. Obtain real-time video surveillance data along the main evacuation route, and identify the personnel density and movement speed in the main evacuation route using the YOLOv5 target detection algorithm;

[0040] S2. When the population density is 1 person / square meter to 3 people / square meter and the moving speed is 0.8 m / s to 1.2 m / s, mark the main evacuation route section as a congested node.

[0041] S3. Retrieve alternative branch paths within a radius of 50-100 meters from the congested node, and calculate the detour distance increment and slope data of the alternative branch paths;

[0042] S4. Select a backup branch path with a detour distance increment of 120-200 meters and a slope of 8-15 degrees as the diversion path;

[0043] S5. Associate the coordinates of the diversion path nodes with the topology of the main evacuation channel nodes to generate a dynamic evacuation road network.

[0044] The evacuation route vector map encapsulation in step 12 includes topological relationship data of the dynamic evacuation road network.

[0045] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, step twelfth further includes: parsing the topological relationship data of the dynamic evacuation road network and extracting the coordinates of the connecting nodes of the diversion paths and the main evacuation channels; deploying LoRa wireless beacons within a radius of 20-50 meters of the connecting nodes, with a beacon broadcast frequency of 1-3 times / second, and the broadcast content including the node code and the path direction angle; after receiving the LoRa beacon data, the application terminal compares it with the locally stored dynamic evacuation road network vector map: if the distance between the current position of the application terminal and the connecting node is 15-30 meters and the direction angle deviation is 20-40 degrees, a voice turning prompt is triggered; if the application terminal's movement trajectory deviates from the planned path of the dynamic evacuation road network for 10-20 seconds, a vibration alarm is activated and the dynamic evacuation road network vector map generated and updated in real time by the cloud platform data processing center is retransmitted to the application terminal; the voice prompt content is generated using a TTS engine to convert the path command, including the turning angle value and a countdown in the range of 15-30 seconds.

[0046] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, step twelfth further includes: when the voice turn prompt is triggered, the gyroscope and accelerometer of the application terminal are activated simultaneously, with a sampling frequency of 50Hz-100Hz; the angular velocity value of the turn action of the application terminal holder is calculated in real time, and the effective turn threshold is set to 15 degrees / second-30 degrees / second;

[0047] If no valid steering action is detected within 3 to 8 seconds after the countdown starts, the remaining countdown will be extended by 5 to 10 seconds; if the cumulative extension reaches 2 to 4 times, the steering voice prompt will be turned off and the vibration alarm will be activated.

[0048] The vibration alarm mode is adjusted according to the type of protective equipment bound to the application terminal: when the protective equipment is a heavy chemical protective suit, a long vibration of 3-5 times for 1-3 seconds is used; when the protective equipment is a light protective suit, a short vibration of 8-12 times for 0.3-0.6 seconds is used; the vibration intensity increases linearly with the ambient noise level in decibels, with a base intensity of 1.5G-2.5G, and an increase of 0.2G-0.5G for every 10-15 decibel increase in noise.

[0049] Preferably, in the intelligent management and control method for safety risks in chemical industrial parks of the present invention, step 12 further includes: real-time acquisition of triaxial acceleration data through the accelerometer of the application terminal, with the sampling frequency maintained at 50Hz-100Hz; when the standard deviation of the Z-axis acceleration exceeds 0.5G-8G for 3-5 consecutive seconds, it is determined that the application terminal is in a state of violent motion.

[0050] During intense activity, the vibration alarm and LED indicator module are activated simultaneously: when the protective equipment is a heavy chemical protective suit, the red light pulse frequency increases to 3Hz-5Hz, and the vibration intensity increases by 0.5G-1.0G; when the protective equipment is a light protective suit, the green light pulse frequency increases to 5Hz-8Hz, the vibration mode switches to intermittent strong vibration, and the working cycle is one vibration every 0.2 seconds to 0.5 seconds, with an interval of 0.1 seconds to 0.3 seconds between each vibration.

[0051] The vibration intensity compensation value is dynamically adjusted according to the motion intensity, including: the reference intensity is the value set during the protective equipment adaptation stage; for every increase of 0.2G-0.4G in the standard deviation of Z-axis acceleration, the vibration intensity increases by 0.1G-0.3G;

[0052] The brightness of the light pulse increases with the smoke concentration. For every 100mg / m³-200mg / m³ increase in smoke concentration, the LED output lumen value increases by 30-80 lumens.

[0053] The present invention has at least the following beneficial effects:

[0054] 1. By establishing a composite judgment rule for leakage through multi-parameter sensor coupling (gas, temperature, pressure, flame), the early warning false alarm rate is effectively reduced; the dual-mode network architecture ensures that key data is uploaded within 5ms, and the integrity rate of pressure change rate, a time-delay sensitive parameter, reaches 99%; dynamic path avoidance of upwind areas and priority connection to fire stations increases the compliance rate of safe evacuation distance to 95% and speeds up rescue response by 40%.

[0055] 2. The differential comparison of LiDAR point cloud enables the identification of newly added obstacles with an accuracy of 50cm, and the path replanning response time is reduced to within 8 seconds; the A-Star algorithm improves the obstacle avoidance path calculation efficiency by 3 times compared with traditional methods, ensuring the real-time availability of updated paths in complex scenarios and reducing the risk of path failure by 70%.

[0056] 3. Terminal priority hierarchical management reduces path update latency for first-priority terminals within 500 meters to below 100ms; Redis queue buffering combined with dynamic bandwidth allocation strategy allocates 60%-80% of bandwidth to first-priority terminals, avoiding network congestion and ensuring 100% instruction arrival rate for terminals in high-risk areas; network bandwidth resources are allocated in a 3:1 ratio to optimize the transmission efficiency of second and third priority terminals, increasing the overall system's concurrent carrying capacity by 2 times.

[0057] 4. Standardized data acquisition process (steps one to five) ensures that the timestamps of multi-source data are aligned, and the delay in leakage judgment is reduced from minutes to within 10 seconds; the automation of path dynamic calculation (step ten) and instruction encapsulation (steps eleven to twelve) improves the efficiency of evacuation plan generation by 5 times; the control latency of 5G module is <800ms, which is 90% faster than traditional manual handling.

[0058] 5. The wind field model generates continuous meteorological vectors through cubic spline interpolation. The angle of the avoidance area is dynamically adjusted with wind speed, expanding from 8m / s-15m / s to 90 degrees, improving the accuracy of gas diffusion coverage prediction by 35%. The path is recalculated every 10-30 seconds to match meteorological changes, increasing the path safety redundancy by 50%.

[0059] 6. Buffer zone overlap detection (40%-70% threshold) enables meteorological vector fusion of multiple leak points, and the composite sector area covers the joint hazard range at 50-120 degrees; the hierarchical planning strategy first constructs the main channel and then derives branch paths, reducing evacuation path conflicts in multi-source leak scenarios by 80% and shortening the overall evacuation time by 25%.

[0060] 7. YOLOv5 identifies people with a density of 1-3 people / m² and a speed of 0.8-1.2m / s, accurately marking congestion nodes; the constraints of 8-15 degree slope and 120-200 meter detour increment ensure the feasibility of diversion paths; dynamic road network topology association improves the traffic efficiency of congested road sections by 40% and avoids the risk of stampedes.

[0061] 8. LoRa beacons (within a radius of 20-50 meters) provide centimeter-level connection node positioning, and a turning prompt is triggered when the directional angle deviation is 20-40 degrees; vibration alarm after 10-20 seconds of deviation from the path + vector map retransmission speeds up the response speed of personnel trajectory correction by 3 times; TTS countdown command reduces the error rate of turning operation by 60%.

[0062] 9. The gyroscope detects effective turning movements of 15-30 degrees per second, enabling interactive verification. The countdown is dynamically extended by 5-10 seconds to adapt to the operation delays of heavy equipment. Differentiated vibration modes are adapted, with long vibrations for heavy equipment and short vibrations for light equipment, increasing the alarm perception rate to 98%. The noise compensation mechanism improves the noise level by 0.2G-0.5G for every 10dB-15dB, ensuring accessibility in high-noise environments.

[0063] 10. The Z-axis acceleration standard deviation of 0.5G-8G accurately determines the state of violent movement; the frequency of red or green light pulses is increased by 3Hz-5Hz or 5Hz-8Hz to enhance the visual capture rate; the vibration intensity is increased by 0.1G-0.3G for every 0.2G-0.4G with motion compensation, so that the alarm recognition rate is maintained above 90% in running conditions; the lumen value increases by 30-80 lumens for every 100mg / m³-200mg / m³ with the increase of smoke concentration, ensuring the visibility of signals in low visibility conditions.

[0064] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the composition structure of an intelligent management and control system for safety risks in chemical industrial parks, as described in one of the technical solutions of this invention. Detailed Implementation

[0066] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0067] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0068] According to one embodiment of the present invention, such as Figure 1 As shown, an intelligent management and control system for safety risks in a chemical industrial park is provided, including sensing layer equipment, data transmission network, cloud platform data processing center and application terminals deployed in the chemical industrial park;

[0069] The sensing layer equipment includes gas concentration sensors, temperature sensors, pressure sensors, and flame detectors installed in the tank area of ​​major hazardous sources;

[0070] The data transmission network adopts a dual-mode transmission architecture of 5G communication module and industrial Ethernet. The sensing layer device packages the collected data and transmits it to the cloud platform data processing center.

[0071] The cloud platform data processing center includes a time-series database, a rule engine, and a risk analysis module. The time-series database stores the collected data from the sensing layer devices in timestamp format and establishes a device code index. The rule engine has preset leakage judgment rules: when the methane concentration data associated with the same device code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data decrease rate is greater than 0.5 MPa / min, a leakage warning is triggered. After receiving the warning signal output by the rule engine, the risk analysis module calls the device coordinate data, real-time meteorological data, and emergency resource distribution data from the 3D geographic information system, and calculates the evacuation path within a radius of 500-2000 meters centered on the leak point using the Dijkstra algorithm. The evacuation path avoids the upwind area of ​​the real-time wind direction and prioritizes connecting the coordinates of the fire station.

[0072] The application terminal receives early warning instructions and evacuation route vector maps issued by the cloud platform data processing center. The early warning instructions include the leakage device code, early warning level, and gas diffusion simulation polygons. The evacuation route vector maps include a dynamically updated set of path node coordinates and the estimated travel time for each node.

[0073] In the sensing layer device of this technical solution, the gas concentration sensor can be configured with a methane detection range of 0%LEL-20%LEL, 0%LEL-50%LEL, or 0%LEL-100%LEL; the temperature sensor can be configured with a detection range of -20℃ to 150℃, -40℃ to 200℃, or -60℃ to 250℃; the pressure sensor can be configured with a range of 0MPa-5MPa, 0MPa-10MPa, or 0MPa-15MPa; and the flame detector can be configured with a response wavelength of 3.3μm-4.3μm, 4.3μm-4.5μm, 4.35μm-4.45μm, or 4.4μm-4.6μm. The 5G communication module of the data transmission network has a latency of ≤20ms, preferably less than 5ms, with selectable thresholds of 1ms, 3ms, or 5ms; the industrial Ethernet transmission bandwidth is 100Mbps-1000Mbps, with selectable thresholds of 100Mbps, 500Mbps, or 1000Mbps. In the leakage detection rules of the rule engine, the methane concentration threshold can be selected as 20% LEL, 25% LEL, or 30% LEL; the continuous sampling period can be selected as 3, 4, or 5 periods; and the pressure drop rate threshold can be selected as 0.3 MPa / min, 0.5 MPa / min, or 0.7 MPa / min. The evacuation path calculation radius can be selected as 500 meters, 1000 meters, or 2000 meters.

[0074] The sensing layer equipment can be commercially available catalytic combustion methane sensors, such as the MC112 model, installed 1.5 meters downwind of the tank's breather valve; PT100 platinum resistance temperature sensors installed in the middle of the tank wall, 2-3 meters above the ground; piezoresistive pressure sensors, such as the MPM4800 model, installed at the pipe flange interface; and infrared flame detectors, such as the IR500 model, installed on the angle steel support of the tank area's dike, 4-6 meters above the ground. The 5G communication module for the data transmission network can be an industrial-grade CPE device deployed in the explosion-proof control cabinet of the tank area; and an industrial Ethernet switch, such as the IES618-2GS model, installed in the park's low-voltage electrical well. The time-series database for the cloud platform data processing center can be InfluxDB deployed on Alibaba Cloud ECS servers; the rule engine uses the Drools rule base running on Kubernetes containers; and the risk analysis module is based on the PostGIS 3D geographic information system and deployed on independent GPU computing nodes. The application terminal can be an explosion-proof tablet computer, such as the T8000 model, equipped at the attachment point of the inspection personnel's protective equipment.

[0075] The working process is as follows: The gas concentration sensor collects methane data every 10 seconds, packages it into JSON format via the Modbus-RTU protocol, including timestamp, device code, and concentration value; temperature and pressure data are sampled at a frequency of 20Hz and uploaded to the cloud platform via a 5G module with a latency of <3ms or an industrial Ethernet with a bandwidth of 500Mbps. The time-series database stores data indexed by device code. The rule engine scans the data stream every 30 seconds. When the methane concentration associated with a certain storage tank code is >25%LEL for four consecutive cycles and the pressure drop rate is >0.5MPa / min, a leak warning is triggered. The risk analysis module calls the real-time wind direction data from the meteorological station (45° northeast wind), calculates an evacuation route with a radius of 1000 meters centered on the leak point, avoids the 60° upwind sector at a distance of 300 meters, and generates the shortest path connecting to the nearest fire station at coordinates X=102.34, Y=35.67 using the Dijkstra algorithm. The application terminal receives early warning commands (including leak tank code K-102, high-risk level, and the coordinate set of the vertex of the gas diffusion polygon) and evacuation route vector diagrams (including 15 path nodes and the passage time of each node). This solution enables multi-parameter collaborative early warning, improving the reliability of leak identification; dual-mode network ensures the integrity of critical data transmission; and dynamic route planning reduces the probability of personnel entering hazardous areas.

[0076] According to another embodiment of the present invention, an intelligent management and control system for safety risks in a chemical industrial park is provided. The risk analysis module further includes a lidar scanning unit, which continuously acquires real-time point cloud data of the path node area after generating the evacuation path. The lidar scanning unit has a scanning frequency of 5Hz-10Hz and a scanning angle range covering 30-60 degrees on both sides of the center line of the evacuation path.

[0077] The risk analysis module performs a differential comparison between real-time point cloud data and the base map elevation data of the 3D geographic information system. When the absolute difference between the surface point cloud elevation value of a newly added obstacle and the base map elevation data is greater than 50 centimeters, the evacuation path replanning is triggered. The replanning process uses the A-Star algorithm to recalculate the obstacle avoidance path, and the updated evacuation path vector map is synchronized to the application terminal within 3 to 8 seconds.

[0078] In this technical solution, the scanning frequency of the lidar scanning unit can be selected as 5Hz, 8Hz, or 10Hz; the scanning angle range can be set to 30 degrees, 45 degrees, or 60 degrees of coverage on one side. For obstacle recognition thresholds, the absolute elevation difference can be selected as 30 cm, 50 cm, or 70 cm; the path replanning response time can be selected as 3 seconds, 5 seconds, or 8 seconds. The path node expansion step size of the A-Star algorithm can be set to 1 meter, 2 meters, or 3 meters; the obstacle avoidance safety distance threshold can be selected as 0.5 meters, 1 meter, or 1.5 meters. The point cloud data sampling interval is 100 milliseconds, 200 milliseconds, or 300 milliseconds. The A-Star algorithm is also known as the A-Star Algorithm.

[0079] The lidar scanning unit can be a 16-line mechanical rotating lidar, installed on top of streetlights in the chemical industrial park, at a height of 8-12 meters above the ground, with an elevation angle adjustable from -5 to +5 degrees. The point cloud processing module can be an embedded industrial computer equipped with a Jetson TX2 chip, deployed in the park's edge computing box, connected to the lidar via an M12 aviation interface. The 3D geographic information system base map is built using the Cesium open-source engine and runs on a cloud platform GPU server. The application terminal can be an explosion-proof handheld terminal equipped with a 5.7-inch display, fixed to the operator's safety belt buckle.

[0080] The working process is as follows: The LiDAR scans the 45-degree area on both sides of the evacuation path centerline at an 8Hz frequency, covering a total of 90 degrees, generating real-time point cloud data, approximately 30,000 points per frame. The risk analysis module performs differential calculations between the point cloud data and pre-stored 3D map elevation data, with a grid resolution of 10cm×10cm. When a grid point cloud elevation value is detected to be 62cm higher than the base map (corresponding to a collapsed pipe rack), it is determined to be a new obstacle. After triggering path replanning, the A-Star algorithm recalculates the path with the leak point as the starting point and the fire station coordinates as the ending point, setting a 1.2-meter obstacle avoidance radius (node ​​expansion step size of 2 meters). The updated path vector map, containing the coordinates of 32 path nodes, is sent to the application terminal via a 5G module within 5 seconds. This solution can achieve centimeter-level obstacle identification, ensuring dynamic path availability; the replanning mechanism reduces the risk of path failure due to sudden environmental changes.

[0081] According to another embodiment of the present invention, in an intelligent management and control system for safety risks in a chemical industrial park, the cloud platform data processing center further includes an instruction distribution module connected to both the risk analysis module and the application terminal. The instruction distribution module includes:

[0082] The terminal positioning unit obtains physical location data through the GPS positioning module built into the application terminal, with a positioning accuracy error of less than 3 meters;

[0083] The distance classification unit calculates the straight-line distance between each application terminal and the leak point, and divides it into three priority levels. The first priority is a distance of less than or equal to 500 meters from the leak point, the second priority is a distance of more than 500 meters but less than or equal to 1000 meters from the leak point, and the third priority is a distance of more than 1000 meters but less than or equal to 2000 meters from the leak point. The warning is automatically lifted when the distance is greater than 2000 meters.

[0084] When evacuation route replanning is triggered, the transmission control unit directly transmits the evacuation route vector map to the first priority application terminal via UDP protocol through a 5G communication module. Data packets transmitted to the second and third priority application terminals are stored in a Redis cache queue and then transmitted via industrial Ethernet after a delay of 2-5 seconds. The resource allocation ratio of the transmission control unit is dynamically adjusted according to the personnel evacuation priority strategy in the emergency plan of the chemical industrial park. The first priority application terminal is allocated 60%-80% of the total bandwidth, and the remaining bandwidth is allocated according to the ratio of the number of second priority application terminals to the number of third priority application terminals = 3:1.

[0085] In this technical solution, the GPS positioning accuracy error threshold of the terminal positioning unit can be set to 1 meter, 2 meters, or 3 meters. In the distance classification unit, the first priority distance threshold can be selected as 300 meters, 500 meters, or 800 meters; the second priority can be selected as 800-1200 meters, 500-1000 meters, or 1000-1500 meters; and the third priority warning cancellation distance threshold can be selected as 1500 meters, 2000 meters, or 2500 meters. The direct transmission delay of the transmission control unit can be selected as 100ms, 200ms, or 300ms; the Redis cache queue delay time can be selected as 2 seconds, 3 seconds, or 5 seconds; the first priority bandwidth allocation can be selected as 60%, 70%, or 80%; and the allocation ratio of the second and third priorities can be selected as 2:1, 3:1, or 4:1. In this technical solution, "resources" specifically refers to the network bandwidth resources managed by the instruction distribution module. The total bandwidth is dynamically allocated by the cloud platform based on the current number of concurrent connections, with a baseline value of 1Gbps-2Gbps. The total resource is the network bandwidth.

[0086] The terminal positioning unit can be a multi-frequency GNSS positioning module, supporting GPS / BeiDou, built into the motherboard of the explosion-proof application terminal, with the antenna located under the plastic shell on the top of the terminal. The distance classification unit can be an edge computing gateway, equipped with an i.MX8 processor, deployed on the second layer of a 19-inch rack in the park's communication equipment room. The Redis cache server of the transmission control unit can be an in-memory database cluster, installed in a cloud platform virtualization environment; the 5G communication module can be an industrial-grade CPE device, fixed 1.5 meters above the ground on an explosion-proof pillar in the tank area; the industrial Ethernet switch can be a 24-port gigabit device, installed in the control center wiring closet. The application terminal shell can be made of flame-retardant PC / ABS alloy material, with the internal circuit board coated with conformal coating.

[0087] The working process is as follows: When the risk analysis module triggers path replanning, the terminal positioning unit obtains the location of a terminal (longitude 118.76°, latitude 32.04°) via GPS. The distance grading unit calculates that the straight-line distance between the terminal and the leak point (longitude 118.75°, latitude 32.05°) is 480 meters, classifying it as the first priority. The transmission control unit directly transmits the path vector map (containing 28 nodes) to the terminal via UDP protocol, with 5G transmission latency controlled within 120ms. Simultaneously, data packets from the second priority terminal (780 meters away) are stored in a Redis queue and sent out via industrial Ethernet after 3 seconds. Bandwidth allocation is dynamically adjusted as follows: the first priority terminal receives 70% (1.4Gbps available bandwidth), and the remaining 30% bandwidth is allocated proportionally according to the number of terminals, with the second priority terminal (15 terminals) receiving 22.5% and the third priority terminal (5 terminals) receiving 7.5%. This solution ensures that terminals in high-risk areas receive path updates first, alleviating network congestion; the tiered latency mechanism optimizes resource utilization.

[0088] According to another embodiment of the present invention, an intelligent management and control method for safety risks in chemical industrial parks is provided, comprising the following steps:

[0089] Step 1: Collect methane concentration data in real time using gas concentration sensors deployed in the tank area of ​​major hazardous sources;

[0090] Step 2: Collect temperature data of the tank area using temperature sensors;

[0091] Step 3: Collect tank area pressure data using pressure sensors;

[0092] Step four: Collect flame radiation data using a flame detector;

[0093] Step 5: The methane concentration data, temperature data, pressure data, and flame radiation data collected in steps 1 to 4 are sent to the cloud platform data processing center via a 5G communication module and an industrial Ethernet dual-mode transmission architecture.

[0094] Step 6: Store the received data in the time-series database of the cloud platform data processing center, and establish a device code index in timestamp format;

[0095] Step 7: Perform leak detection through the rule engine: When the methane concentration data associated with the same equipment code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data drop rate is greater than 0.5 MPa / min, a leak warning signal is generated.

[0096] Step 8: Respond to the leak warning signal and call up the equipment coordinate data, real-time meteorological data, and emergency resource distribution data stored in the 3D geographic information system;

[0097] Step 9: Based on real-time wind direction data, set the evacuation route avoidance area. The evacuation route area covers a fan-shaped area 30-60 degrees upwind of the leak point and is 200-800 meters away from the leak point.

[0098] Step 10: Calculate evacuation routes within a radius of 500-2000 meters centered on the leak point using the Dijkstra algorithm. The evacuation routes will bypass the evacuation route avoidance area set in Step 9 and will preferentially connect to the coordinates of the nearest fire station.

[0099] Step 11: Encapsulate the leaking device code, warning level, and gas diffusion simulation polygon into a warning command;

[0100] Step 12: Encapsulate the coordinate set of evacuation route nodes, the estimated travel time of each node, and the real-time wind direction vector into an evacuation route vector map.

[0101] Step 13: Send early warning commands and evacuation route vector maps to application terminals via the 5G communication module.

[0102] In step one of this technical solution, the methane concentration detection range can be selected from 0%LEL-50%LEL, 0%LEL-100%LEL, or 0%LEL-150%LEL. In step two, the temperature detection range can be selected from -30℃ to 150℃, -40℃ to 200℃, or -50℃ to 250℃. In step three, the pressure detection range can be selected from 0MPa-8MPa, 0MPa-10MPa, or 0MPa-12MPa. In step four, the flame detector response wavelength can be selected from 3.3μm-4.3μm, 4.3μm-4.5μm, 4.35μm-4.45μm, or 4.4μm-4.6μm. In step five, the 5G transmission latency can be selected from 3ms, 5ms, or 8ms; the industrial Ethernet bandwidth can be selected from 100Mbps, 500Mbps, or 1000Mbps. Step 7: Leakage detection continuous cycle can be selected as 3, 4, or 5 cycles; methane threshold can be selected as 20%, 25%, or 30% LEL; pressure drop rate threshold can be selected as 0.4 MPa / min, 0.5 MPa / min, or 0.6 MPa / min. Step 9: Avoidance zone angle can be selected as 30 degrees, 45 degrees, or 60 degrees; distance can be selected as 200 meters, 500 meters, or 800 meters. Step 10: Evacuation radius can be selected as 500 meters, 1000 meters, or 2000 meters. Step 13: Distribution delay can be selected as 500ms, 800ms, or 1000ms.

[0103] The sensors used in steps one through four can be the same models used in the aforementioned technical solutions: a catalytic combustion methane sensor is installed 1.5 meters downwind of the tank's breather valve at the flange interface; a PT100 temperature sensor is mounted on a bracket 2 meters above the ground on the tank side wall; a piezoresistive pressure sensor is connected to the drain valve interface of the pipeline; and an infrared flame detector is fixed 4.5 meters above the ground on the angle steel column of the tank area's dike. The 5G communication module in step five can be an industrial-grade CPE deployed inside the explosion-proof control box in the tank area; the industrial Ethernet switch is installed in the low-voltage room cabinet in the park. The time-series database in step six can be InfluxDB running on an Alibaba Cloud 4-core 8GB ECS server. The 3D geographic information system in step eight can be the SuperMap platform deployed on a dedicated GPU server. The application terminal can be an explosion-proof tablet computer, equipped with the operator's safety belt. Step eight also includes calculating the gas diffusion range based on the Gaussian plume diffusion model. The input parameters include: the coordinates of the leak point, real-time wind speed data, wind direction angle data, and the duration of the leak. The output is the diffusion concentration contour lines. The closed regions of the contour lines with a concentration ≥ 25% LEL are converted into polygon vertex coordinates and stored as gas diffusion simulation polygons.

[0104] The working process is as follows: Temperature sensors collect tank wall temperature data every 15 seconds (current value 65℃), and pressure sensors monitor pipeline pressure at a frequency of 20Hz (current value 3.2MPa). The data is packaged into JSON format via the Modbus-RTU protocol. It is then uploaded to the cloud platform via a 5G module with a latency of 4ms. The time-series database stores the data according to the device code and establishes an index. The rule engine scans the data every 30 seconds. When the methane concentration associated with the code B-203 of the storage tank is >25%LEL for four consecutive times and the pressure drop rate is >0.5MPa / min, a leak warning is triggered. The risk analysis module calls real-time northeast wind data to delineate a 45-degree upwind avoidance zone centered on the leak point, with a distance of 500 meters. An evacuation route with a radius of 1000 meters is generated using the Dijkstra algorithm (bypassing the avoidance zone and connecting to the fire station's X102 / Y58 coordinates). The early warning command includes a leak code, high-risk level, and coordinates of the gas diffusion polygon; the evacuation route vector map includes 18 nodes and the passage time at each node, and is sent to the terminal within 720ms via a 5G module. This solution automates leak detection, improving emergency response speed; dynamic route avoidance reduces the risk of personnel exposure.

[0105] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, step eight further includes: extracting wind speed data and wind direction angle data from real-time meteorological data, wherein the wind speed detection range is 1 m / s to 30 m / s and the wind direction angle detection range is 0 degrees to 360 degrees; performing cubic spline interpolation calculation on the wind speed data and wind direction angle data with an interpolation period of 15 seconds to 40 seconds to generate a continuous time series wind field model; and associating the output of the wind field model with the equipment code of the leakage point and storing it as dynamic meteorological vector data.

[0106] The path avoidance area set in step nine is updated based on dynamic meteorological vector data. When the wind speed data is 8 m / s-15 m / s, the angle of the fan-shaped area expands to the range of 45 degrees-90 degrees.

[0107] In step ten, the calculation frequency of evacuation routes is increased to once every 10-30 seconds.

[0108] In step eight of this technical solution, the wind speed detection range can be selected as 1m / s-10m / s, 1m / s-20m / s, or 1m / s-30m / s; the wind direction angle detection adopts 0°-360° full coverage. The interpolation period can be selected as 15 seconds, 25 seconds, or 40 seconds; the number of cubic spline interpolation nodes can be set to 4 points, 6 points, or 8 points. In step nine, the wind speed trigger value in the avoidance area angle adjustment threshold can be selected as 6m / s, 8m / s, or 10m / s; the angle expansion range can be selected as 45°-60°, 60°-75°, or 75°-90°. In step ten, the path calculation frequency can be selected to be executed every 10 seconds, 20 seconds, or 30 seconds.

[0109] The wind speed and direction sensor can be an ultrasonic weather station, installed on the rooftop lightning rod base of the tallest building in the chemical industrial park, at a height of no less than 20 meters above the ground. The interpolation calculation module can be a cloud platform GPU server equipped with an NVIDIA T4 computing card, deployed on the third layer of a data center rack. The dynamic meteorological vector storage unit can be a Redis time-series database, running in a 16GB memory container. The 3D geographic information system is implemented using the SuperMap platform software, connected to the meteorological data interface via TCP / IP protocol. The application terminal display can be an explosion-proof LCD panel made of Gorilla Glass, embedded in the terminal casing.

[0110] The working process is as follows: The ultrasonic weather station collects wind speed (current value 12 m / s) and wind direction (current value 135°) every 2 seconds and uploads them to the cloud platform via the OPC protocol. With an interpolation period of 25 seconds, cubic spline interpolation is performed on the wind speed and direction data from the most recent four periods to generate a continuous wind field model (including wind speed vectors at 5° intervals in the 0°-360° direction). When the wind speed value is consistently >8 m / s for 30 seconds, step nine expands the avoidance area angle from 60° to 80°. The risk analysis module calls the updated dynamic meteorological vector every 20 seconds to recalculate the evacuation path (if the original path node N15 is located within the diffusion zone, it is adjusted to the downwind safe node N16). The path vector map includes wind direction arrows and is distributed via the 5G module. This solution dynamically adjusts the avoidance area according to weather changes, improving path safety; high-frequency recalculation adapts to sudden weather changes.

[0111] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, step eight further includes: retrieving the equipment codes associated with all current leak warning signals, extracting the coordinate data of each leak point and dynamic meteorological vector data; establishing a buffer zone with a radius of 30-60 meters centered on the leak point coordinates, and using a spatial overlay analysis algorithm to detect the overlapping part of the buffer zone; when the overlapping area accounts for more than 40%-70%, merging the dynamic meteorological vector data and generating a joint path avoidance area;

[0112] In step nine, the range of the path avoidance area is expanded to: a 30-60 degree sector area for a single leak point; and a 50-120 degree composite sector area for a combined path avoidance area, with the distance taken as the maximum range of 200-800 meters among all leak points.

[0113] In step ten, the evacuation route calculation adopts a hierarchical planning strategy: first, the main evacuation route is planned, connecting the coordinates of the safety exits within a radius of 500-2000 meters; then, branch routes are generated based on the main evacuation route to the coding areas of each equipment.

[0114] In step eight of this technical solution, the radius of the buffer zone can be selected as 30 meters, 45 meters, or 60 meters; the threshold for the overlap area ratio can be selected as 40%, 55%, or 70%. In step nine, the angle of the composite sector can be selected as 50°-80°, 80°-100°, or 100°-120°; the distance range is taken as the maximum of 200 meters-400 meters, 400 meters-600 meters, or 600 meters-800 meters for each leakage point. In the ten-layer planning step, the connection radius of the main channel safety exit can be selected as 500 meters-1000 meters, 1000 meters-1500 meters, or 1500 meters-2000 meters; the granularity of branch path generation can be selected as 50 meters, 100 meters, or 150 meters grid.

[0115] The spatial overlay analysis module can use ArcGIS Engine components, deployed on a cloud platform spatial computing server. The buffer area generation unit can use a PostGIS geodatabase extension, running on a 128GB memory database node. The dynamic meteorological vector storage uses a Redis time-series database. The main channel planning server can use a high-availability cluster (dual Xeon Gold processors), installed on the 5th floor of the data center rack. Video surveillance equipment can use explosion-proof dome cameras with stainless steel casings, mounted on a bracket 6 meters above the ground in the tank area pipe gallery. The merging rules for dynamic meteorological vector data can be as follows: take the maximum wind speed among all leak points as the joint wind speed, and take the wind direction of the leak point with the maximum wind speed as the dominant wind direction; if the leak point with the maximum wind speed is not unique, then take the wind direction of the geometric center point.

[0116] The working process is as follows: When both tank area T-201 (coordinates X=102.35, Y=35.68) and distillation tower R-305 (coordinates X=102.41, Y=35.72) trigger a leak warning simultaneously, step eight establishes a 50-meter radius buffer circle centered on each point. Spatial overlay analysis detects that the overlap area of ​​the two circles reaches 58% (>55% threshold), and dynamic meteorological vectors are merged to generate a joint avoidance area. Step nine sets the angle of the composite sector area to 100°, and the distance is set to the maximum value of 600 meters (covering the 180-meter gap between the two leak points). Step ten first plans the main evacuation channel connecting to the No. 2 safety exit on the west side (800 meters from T-201), and then generates branch paths to the T-201 tank area operating platform (containing 3 path nodes) and the R-305 control room (containing 2 path nodes). This scheme can handle the risk of multiple leak point coupling and avoid path intersection conflicts; hierarchical planning optimizes calculation efficiency.

[0117] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, the branch path generation stage in step ten includes the following steps:

[0118] S1. Obtain real-time video surveillance data along the main evacuation route, and identify the personnel density and movement speed in the main evacuation route using the YOLOv5 target detection algorithm;

[0119] S2. When the population density is 1 person / square meter to 3 people / square meter and the moving speed is 0.8 m / s to 1.2 m / s, mark the main evacuation route section as a congested node.

[0120] S3. Retrieve alternative branch paths within a radius of 50-100 meters from the congested node, and calculate the detour distance increment and slope data of the alternative branch paths;

[0121] S4. Select a backup branch path with a detour distance increment of 120-200 meters and a slope of 8-15 degrees as the diversion path;

[0122] S5. Associate the coordinates of the diversion path nodes with the topology of the main evacuation channel nodes to generate a dynamic evacuation road network.

[0123] The evacuation route vector map encapsulation in step 12 includes topological relationship data of the dynamic evacuation road network.

[0124] In step S1 of this technical solution, the personnel density detection threshold can be selected as 1 person / m², 2 people / m², or 3 people / m²; the movement speed threshold can be selected as 0.8 m / s, 1.0 m / s, or 1.2 m / s. The congestion node marking conditions in step S2 are: density of 1-2 people / m² and speed of 0.8-1.0 m / s, or density of 2-3 people / m² and speed of 1.0-1.2 m / s. In step S3, the alternative route retrieval radius can be selected as 50 meters, 75 meters, or 100 meters; the detour distance increment threshold is set to 120 meters, 160 meters, or 200 meters; and the slope threshold can be selected as 8°, 12°, or 15°. The diversion path selection constraint in step S4 is that both the detour increment and slope must be ≤180 meters and ≤12° simultaneously.

[0125] The video surveillance equipment can be a 2-megapixel explosion-proof network camera with a 316L stainless steel casing or a 4K resolution camera, installed on the pipe gallery support above the main evacuation route, at a height not exceeding 6 meters above the ground and a 15° downward angle. The target detection server can be an edge computing box equipped with a Jetson Xavier NX chip, deployed in an explosion-proof control box on the side wall of the route. The slope detection sensor can be a MEMS inclinometer (range ±30°), fixed to the back of the branch path sign. The dynamic evacuation network generation module is implemented using the ArcGIS Engine spatial analysis component and runs on a cloud platform spatial analysis server. The application terminal display can be a high-brightness IPS LCD panel with anti-reflective coating, embedded in the front of the terminal casing.

[0126] The working process is as follows: When the video surveillance of the main passage C area shows a personnel density of 2.3 people / square meter, and the YOLOv5 algorithm identifies the average moving speed as 0.9 m / s, step S2 marks this section as a congested node. Step S3 retrieves three alternative branch paths within an 80-meter radius of the node: Path A has a detour increment of 150 meters / slope of 10°, Path B has a detour increment of 220 meters / slope of 8°, and Path C has a detour increment of 180 meters / slope of 16°. Step S4 selects Path A (satisfying an increment ≤ 180 meters and a slope ≤ 12°). Step S5 topologically associates the nodes of Path A (N45-N48) with the main passage node N32, generating dynamic evacuation road network vector data. The application terminal receives the updated path map and displays the diversion arrows. This solution can alleviate congestion on the main passage and ensure smooth evacuation; the slope constraint ensures path feasibility.

[0127] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, step twelf further includes: parsing the topological relationship data of the dynamic evacuation road network and extracting the coordinates of the connecting nodes between the diversion paths and the main evacuation channels; deploying LoRa wireless beacons within a radius of 20-50 meters of the connecting nodes, with a beacon broadcast frequency of 1-3 times / second, and the broadcast content including the node code and the path direction angle; after receiving the LoRa beacon data, the application terminal compares it with the locally stored dynamic evacuation road network vector map: if the distance between the current position of the application terminal and the connecting node is 15-30 meters and the direction angle deviation is 20-40 degrees, a voice turning prompt is triggered; if the application terminal's movement trajectory deviates from the planned path of the dynamic evacuation road network for 10-20 seconds, a vibration alarm is activated and the dynamic evacuation road network vector map generated and updated in real time by the cloud platform data processing center is retransmitted to the application terminal; the voice prompt content is generated using a TTS engine to convert the path command, including the turning angle value and a countdown in the range of 15-30 seconds.

[0128] In this technical solution, the deployment radius of the connecting nodes can be selected as 20 meters, 35 meters, or 50 meters; the beacon broadcast frequency can be selected as 1, 2, or 3 times / second. The directional angle deviation threshold is set to 20°, 30°, or 40°; the deviation from the path duration threshold can be selected as 10 seconds, 15 seconds, or 20 seconds. The TTS countdown range can be selected as 15 seconds, 22 seconds, or 30 seconds. The accuracy of the voice prompt turning angle value is ±5°.

[0129] The LoRa wireless beacon can be an industrial-grade low-power module (IP68 housing), installed in the middle of a streetlight pole at the junction of a branch path and the main road, 2.5 meters above the ground. The gyroscope module can be a MEMS six-axis sensor (range ±2000° / sec), integrated into the application terminal's motherboard. The accelerometer uses a three-axis MEMS sensor. The TTS engine can be an offline speech synthesis chip supporting Chinese commands, integrated next to the terminal processor. The vibration motor can be a flat eccentric rotor motor (10mm diameter), fixed inside the terminal housing.

[0130] The working process is as follows: When the application terminal approaches the connection node (coordinates X=102.56, Y=35.73), the LoRa beacon broadcasts the node code N58 and the azimuth angle 285° at a frequency of 2Hz. The terminal compares its local path map and detects that its current position is 25 meters away from the node and its azimuth angle deviation is 35° (>30° threshold), triggering a voice prompt: "Turn 40 degrees to the left, countdown 20 seconds." If the terminal's movement trajectory deviates from the planned path for 15 seconds (such as zigzagging), a vibration alarm is activated and the updated vector map (including the newly added offshoot node N59) is retransmitted. The vibration intensity increases to 2.2G as the ambient noise (currently 85dB) increases. This solution achieves centimeter-level path correction, reducing the risk of personnel getting lost; multimodal alarms adapt to complex environments.

[0131] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, step twelf further includes: when the voice turn prompt is triggered, the gyroscope and accelerometer of the application terminal are activated simultaneously, with a sampling frequency of 50Hz-100Hz; the angular velocity value of the turn action of the application terminal holder is calculated in real time, and the effective turn threshold is set to 15 degrees / second-30 degrees / second.

[0132] If no valid steering action is detected within 3 to 8 seconds after the countdown starts, the remaining countdown will be extended by 5 to 10 seconds; if the cumulative extension reaches 2 to 4 times, the steering voice prompt will be turned off and the vibration alarm will be activated.

[0133] The vibration alarm mode is adjusted according to the type of protective equipment bound to the application terminal: when the protective equipment is a heavy chemical protective suit, a long vibration of 3-5 times for 1-3 seconds is used; when the protective equipment is a light protective suit, a short vibration of 8-12 times for 0.3-0.6 seconds is used; the vibration intensity increases linearly with the ambient noise level in decibels, with a base intensity of 1.5G-2.5G, and an increase of 0.2G-0.5G for every 10-15 decibel increase in noise.

[0134] The gyroscope sampling frequency in this technical solution can be selected as 50Hz, 75Hz, or 100Hz; the effective turning threshold is set to 15° / second, 22° / second, or 30° / second. The initial countdown extension can be selected as 5 seconds, 7 seconds, or 10 seconds; the cumulative extension threshold is set to 2, 3, or 4 times. The long vibration mode can be selected as 3 vibrations lasting 1.5 seconds, 4 vibrations lasting 2 seconds, or 5 vibrations lasting 3 seconds; the short vibration mode can be selected as 8 vibrations lasting 0.4 seconds, 10 vibrations lasting 0.5 seconds, or 12 vibrations lasting 0.6 seconds. The noise compensation reference strength can be selected as 1.5G, 2.0G, or 2.5G; the noise is increased by 0.2G, 0.3G, or 0.5G per 10dB.

[0135] The gyroscope module can be a MEMS angular velocity sensor (±4000dps range), soldered onto the terminal PCB along with the accelerometer. The vibration motor can be a linear resonator (LRA) or an eccentric rotor motor (ERM), glued to the side wall of the terminal's battery compartment. The protective equipment mounting unit can be an anti-metal NFC tag, sewn onto the inside of the right shoulder strap of the protective suit. The noise sensor can be a digital microphone module (frequency response 100Hz-10kHz), embedded in the opening at the top of the terminal.

[0136] The working process is as follows: After receiving a turning prompt, the operator wearing heavy protective clothing uses a gyroscope to sample and detect the angular velocity at 80Hz (current value 18° / s < 22° / s threshold). If the turning threshold is not reached after a 5-second countdown, the countdown is extended by 7 seconds. If a valid turn is still not detected after 3 extensions, the voice prompt is turned off and a long vibration mode is activated (4 times for 2 seconds each, intensity 2.8G). The vibration intensity increases to 3.3G (baseline 2.5G + 0.8G compensation) as the compressor noise (95dB) increases. This solution adapts to the operation delay of heavy equipment, ensuring that critical commands are received; the intensity compensation mechanism maintains alarm awareness.

[0137] According to another embodiment of the present invention, in the intelligent management and control method for safety risks in chemical industrial parks, step 12 further includes: real-time acquisition of triaxial acceleration data through the accelerometer of the application terminal, with the sampling frequency maintained at 50Hz-100Hz; when the standard deviation of the Z-axis acceleration exceeds 0.5G-8G for 3-5 consecutive seconds, it is determined that the application terminal is in a state of violent motion.

[0138] During intense activity, the vibration alarm and LED indicator module are activated simultaneously: when the protective equipment is a heavy chemical protective suit, the red light pulse frequency increases to 3Hz-5Hz, and the vibration intensity increases by 0.5G-1.0G; when the protective equipment is a light protective suit, the green light pulse frequency increases to 5Hz-8Hz, the vibration mode switches to intermittent strong vibration, and the working cycle is one vibration every 0.2 seconds to 0.5 seconds, with an interval of 0.1 seconds to 0.3 seconds between each vibration.

[0139] The vibration intensity compensation value is dynamically adjusted according to the motion intensity, including: the reference intensity is the value set during the protective equipment adaptation stage; for every increase of 0.2G-0.4G in the standard deviation of Z-axis acceleration, the vibration intensity increases by 0.1G-0.3G;

[0140] The brightness of the light pulse increases with the smoke concentration. For every 100mg / m³-200mg / m³ increase in smoke concentration, the LED output lumen value increases by 30-80 lumens.

[0141] The accelerometer sampling frequency in this technical solution is maintained at 50Hz, 75Hz, or 100Hz. The threshold for determining severe motion is: the standard deviation of Z-axis acceleration can be selected as 0.5G, 4.0G, or 8.0G; the continuous duration can be selected as 3 seconds, 4 seconds, or 5 seconds. The red light pulse frequency can be selected as 3Hz, 4Hz, or 5Hz; the green light pulse frequency can be selected as 5Hz, 6Hz, or 8Hz. Vibration intensity compensation value: 0.1G increase for every 0.2G standard deviation, 0.2G increase for every 0.3G, or 0.3G increase for every 0.4G. Smoke compensation: 30 lumens increase for every 100mg / m³, 55 lumens increase for every 150mg / m³, or 80 lumens increase for every 200mg / m³.

[0142] The triaxial accelerometer can be an industrial-grade MEMS module (±16g range), mounted on the center of the terminal motherboard. The LED indicator module can be a high-brightness RGB LED (200 lumens), embedded under the transparent protective cover on the top of the terminal. The smoke sensor can be a laser scattering probe (range 0mg / m³-500mg / m³), installed behind the side air intake grille of the terminal. The vibration motor is an eccentric rotor motor, connected in parallel with the LED light synchronization control circuit.

[0143] The working process is as follows: When a person is running, the accelerometer samples at 90Hz. If the Z-axis standard deviation reaches 6.2G (>4.0G threshold) for 4 consecutive seconds, a state of intense motion is determined. When wearing light protective clothing: a green light pulse (6Hz frequency) is activated and the intermittent strong vibration mode is switched (vibration once every 0.3 seconds / 0.2-second interval). The vibration intensity is increased from the baseline 1.8G to 2.4G (compensation value = 6.2G / 0.3×0.15). At the same time, the ambient smoke concentration is detected at 220mg / m³, and the LED lumen value is increased from 150 lumens to 260 lumens (baseline value + 110 lumen compensation). This solution enhances the alarm recognition rate under motion conditions; the light-vibration linkage compensation mechanism improves the reliability of perception in harsh environments.

[0144] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0145] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. An intelligent management and control system for safety risks in chemical industrial parks, characterized in that: This includes sensing layer equipment, data transmission networks, cloud platform data processing centers, and application terminals deployed within the chemical industrial park; The sensing layer equipment includes gas concentration sensors, temperature sensors, pressure sensors, and flame detectors installed in the tank area of ​​major hazardous sources; The data transmission network adopts a dual-mode transmission architecture of 5G communication module and industrial Ethernet. The sensing layer device packages the collected data and transmits it to the cloud platform data processing center. The cloud platform data processing center includes a time-series database, a rule engine, and a risk analysis module. The time-series database stores the collected data from the sensing layer devices in timestamp format and establishes a device code index. The rule engine has pre-set leakage judgment rules: when the methane concentration data associated with the same device code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data decrease rate is greater than 0.5 MPa / min, a leakage warning is triggered. After receiving the warning signal output by the rule engine, the risk analysis module calls the device coordinate data, real-time meteorological data, and emergency resource distribution data from the 3D geographic information system. It extracts wind speed data and wind direction angle data from the real-time meteorological data. The wind speed detection range is 1-30 m / s, and the wind direction angle detection range is 0-360 degrees. With an interpolation period of 15-40 seconds, cubic spline interpolation calculation is performed on the wind speed data and wind direction angle data to generate a continuous time series wind field model. The wind field model output is associated with the device code of the leakage point and stored as dynamic meteorological vector data. The system retrieves the device codes associated with all current leak warning signals and extracts the coordinate data and dynamic meteorological vector data of each leak point. A buffer zone with a radius of 30-60 meters is established centered on the leak point coordinates, and a spatial overlay analysis algorithm is used to detect overlapping areas within the buffer zone. When the overlapping area exceeds 40%-70%, the dynamic meteorological vector data is merged to generate a joint path avoidance area. An evacuation path avoidance area is set based on real-time wind direction data, covering a 30-60 degree upwind sector from the leak point and extending 200-800 meters from the leak point. The set path avoidance area is updated based on dynamic meteorological vector data; when the wind speed is 8-15 m / s, the sector angle is expanded to 45-90 degrees. The Dijkstra algorithm is used to calculate evacuation paths within a radius of 500-2000 meters centered on the leak point. These evacuation paths avoid the upwind area of ​​the real-time wind direction and prioritize connections to fire station coordinates. The calculation frequency of the evacuation paths is increased to once every 10-30 seconds. The application terminal receives early warning instructions and evacuation route vector maps issued by the cloud platform data processing center. The early warning instructions include the leakage device code, early warning level, and gas diffusion simulation polygon. The evacuation route vector map includes a dynamically updated set of path node coordinates and the estimated travel time for each node. The risk analysis module also includes a lidar scanning unit, which continuously acquires real-time point cloud data of the path node areas after the evacuation path is generated; the lidar scanning unit has a scanning frequency of 5-10Hz and a scanning angle range covering 30-60 degrees on both sides of the center line of the evacuation path. The risk analysis module performs a differential comparison between real-time point cloud data and the base map elevation data of the 3D geographic information system. When the absolute difference between the surface point cloud elevation value of a newly added obstacle and the base map elevation data is greater than 50 centimeters, the evacuation route replanning is triggered. The replanning process uses the A* algorithm to recalculate the obstacle avoidance path, and the updated evacuation route vector map is synchronized to the application terminal within 3-8 seconds. The cloud platform data processing center also includes an instruction distribution module connected to both the risk analysis module and the application terminal. The instruction distribution module includes: The terminal positioning unit obtains physical location data through the GPS positioning module built into the application terminal, with a positioning accuracy error of less than 3 meters; The distance classification unit calculates the straight-line distance between each application terminal and the leak point, and divides it into three priority levels. The first priority is a distance of less than or equal to 500 meters from the leak point, the second priority is a distance of more than 500 meters but less than or equal to 1000 meters from the leak point, and the third priority is a distance of more than 1000 meters but less than or equal to 2000 meters from the leak point. The warning is automatically lifted when the distance is greater than 2000 meters. When evacuation route replanning is triggered, the transmission control unit directly transmits the evacuation route vector map to the first priority application terminal via UDP protocol through a 5G communication module. Data packets transmitted to the second and third priority application terminals are stored in a Redis cache queue and then transmitted via industrial Ethernet after a delay of 2-5 seconds. The resource allocation ratio of the transmission control unit is dynamically adjusted according to the personnel evacuation priority strategy in the emergency plan of the chemical industrial park. The first priority application terminal is allocated 60%-80% of the total bandwidth, and the remaining bandwidth is allocated according to the ratio of the number of second priority application terminals to the number of third priority application terminals = 3:

1.

2. A method for intelligent management and control of safety risks in chemical industrial parks, characterized in that: Includes the following steps: Step 1: Collect methane concentration data in real time using gas concentration sensors deployed in the tank area of ​​major hazardous sources; Step 2: Collect temperature data of the tank area using temperature sensors; Step 3: Collect tank area pressure data using pressure sensors; Step four: Collect flame radiation data using a flame detector; Step 5: The methane concentration data, temperature data, pressure data, and flame radiation data collected in steps 1 to 4 are sent to the cloud platform data processing center via a 5G communication module and an industrial Ethernet dual-mode transmission architecture. Step 6: Store the received data in the time-series database of the cloud platform data processing center, and establish a device code index in timestamp format; Step 7: Perform leak detection through the rule engine: When the methane concentration data associated with the same equipment code exceeds 25% LEL for 3-5 consecutive sampling cycles, and the pressure data drop rate is greater than 0.5 MPa / min, a leak warning signal is generated. Step 8: Respond to the leak warning signal and call the equipment coordinate data, real-time meteorological data, and emergency resource distribution data stored in the 3D geographic information system; extract wind speed and wind direction angle data from the real-time meteorological data, with a wind speed detection range of 1-30 m / s and a wind direction angle detection range of 0-360 degrees; perform cubic spline interpolation calculations on the wind speed and wind direction angle data with an interpolation period of 15-40 seconds to generate a continuous time series wind field model; associate the wind field model output with the equipment code of the leak point and store it as dynamic meteorological vector data; retrieve the equipment codes associated with all current leak warning signals and extract the coordinate data and dynamic meteorological vector data of each leak point; establish a buffer zone with a radius of 30-60 meters centered on the leak point coordinates, and use a spatial overlay analysis algorithm to detect the overlapping parts of the buffer zone; when the overlapping area exceeds 40%-70%, merge the dynamic meteorological vector data and generate a joint path avoidance area; Step 9: Based on real-time wind direction data, set the evacuation route avoidance area. The evacuation route area covers a fan-shaped area 30-60 degrees upwind of the leak point and is 200-800 meters away from the leak point. The set route avoidance area is updated based on dynamic meteorological vector data. When the wind speed data is 8-15 m / s, the angle of the fan-shaped area is expanded to 45-90 degrees. Step 10: Calculate evacuation routes within a radius of 500-2000 meters centered on the leak point using the Dijkstra algorithm. The evacuation routes will bypass the evacuation route avoidance area set in Step 9 and will preferentially connect to the nearest fire station coordinates. The calculation frequency of the evacuation routes will be increased to once every 10-30 seconds. Step 11: Encapsulate the leaking device code, warning level, and gas diffusion simulation polygon into a warning command; Step 12: Encapsulate the coordinate set of evacuation route nodes, the estimated travel time of each node, and the real-time wind direction vector into an evacuation route vector map. Step 13: Send early warning commands and evacuation route vector maps to application terminals via the 5G communication module.

3. The intelligent management and control method for safety risks in chemical industrial parks as described in claim 2, characterized in that, In step nine, the range of the path avoidance area is expanded to: a 30-60 degree sector area for a single leak point; and a 50-120 degree composite sector area for a combined path avoidance area, with the distance taken as the maximum range of 200-800 meters among all leak points. In step ten, the evacuation route calculation adopts a hierarchical planning strategy: first, the main evacuation routes are planned, connecting the coordinates of safe exits within a radius of 500-2000 meters; Then, branch paths are generated based on the main evacuation channels to the coding areas of each device.

4. The intelligent management and control method for safety risks in chemical industrial parks as described in claim 3, characterized in that, Step 10, the branch path generation stage, includes the following steps: S1. Obtain real-time video surveillance data along the main evacuation route, and identify the personnel density and movement speed in the main evacuation route using the YOLOv5 target detection algorithm; S2. When the population density is 1-3 people / square meter and the movement speed is 0.8-1.2 meters / second, mark the main evacuation route section as a congested node. S3. Retrieve alternative branch paths within a radius of 50-100 meters from the congested node, and calculate the detour distance increment and slope data of the alternative branch paths; S4. Select a backup branch path with an incremental detour distance of 120-200 meters and a gradient of 8-15 degrees as the diversion path; S5. Associate the coordinates of the diversion path nodes with the topology of the main evacuation channel nodes to generate a dynamic evacuation road network. The evacuation route vector map encapsulation in step 12 includes topological relationship data of the dynamic evacuation road network.

5. The intelligent management and control method for safety risks in chemical industrial parks as described in claim 4, characterized in that, Step 12 further includes: parsing the topological relationship data of the dynamic evacuation road network and extracting the coordinates of the connecting nodes between the diversion paths and the main evacuation channels; deploying LoRa wireless beacons within a radius of 20-50 meters of the connecting nodes, with a beacon broadcast frequency of 1-3 times / second, and the broadcast content including the node code and path direction angle; after receiving the LoRa beacon data, the application terminal compares it with the locally stored dynamic evacuation road network vector map: if the distance between the current position of the application terminal and the connecting node is 15-30 meters and the direction angle deviation is 20-40 degrees, a voice turning prompt is triggered; if the application terminal's movement trajectory deviates from the planned path of the dynamic evacuation road network for 10-20 seconds, a vibration alarm is activated and the dynamic evacuation road network vector map generated and updated in real time by the cloud platform data processing center is retransmitted to the application terminal; the voice prompt content is generated using a TTS engine to convert path instructions, including the turning angle value and a countdown in the range of 15-30 seconds.

6. The intelligent management and control method for safety risks in chemical industrial parks as described in claim 5, characterized in that, Step 12 also includes: when the voice direction prompt is triggered, the gyroscope and accelerometer of the application terminal are activated simultaneously, with a sampling frequency of 50-100Hz; the angular velocity value of the application terminal holder's turning action is calculated in real time, and the effective turning threshold is set to 15-30 degrees / second; If no valid steering action is detected within 3-8 seconds after the countdown starts, the remaining countdown will be extended by 5-10 seconds; if the cumulative extension reaches 2-4 times, the steering voice prompt will be turned off and the vibration alarm will be activated. The vibration alarm mode is adjusted according to the type of protective equipment bound to the application terminal: when the protective equipment is a heavy chemical protective suit, it uses 3-5 long vibrations lasting 1-3 seconds; when the protective equipment is a light protective suit, it uses 8-12 short vibrations lasting 0.3-0.6 seconds. The vibration intensity increases linearly with the ambient noise level in decibels, with a base intensity of 1.5-2.5G. For every 10-15 decibel increase in noise, the intensity increases by 0.2-0.5G.

7. The intelligent management and control method for safety risks in chemical industrial parks as described in claim 6, characterized in that, Step 12 also includes: collecting triaxial acceleration data in real time through the accelerometer of the application terminal, with a sampling frequency of 50-100Hz; when the standard deviation of the Z-axis acceleration exceeds 0.5-8G for 3-5 consecutive seconds, the application terminal is determined to be in a state of violent motion. During intense activity, the vibration alarm and LED indicator module are activated simultaneously: when the protective equipment is a heavy chemical protective suit, the red light pulse frequency increases to 3-5Hz, and the vibration intensity increases by 0.5-1.0G; when the protective equipment is a light protective suit, the green light pulse frequency increases to 5-8Hz, the vibration mode switches to intermittent strong vibration, and the working cycle is one vibration every 0.2 to 0.5 seconds, with an interval of 0.1 to 0.3 seconds between each vibration. The vibration intensity compensation value is dynamically adjusted according to the motion intensity, including: the reference intensity is the value set during the adaptation stage of the protective equipment; for every increase of 0.2-0.4G in the standard deviation of Z-axis acceleration, the vibration intensity increases by 0.1-0.3G; The brightness of the light pulse increases with the smoke concentration. For every 100-200 mg / m³ increase in smoke concentration, the LED output lumen value increases by 30-80 lumens.

Images