Reservoir dam flood discharge influence river reach safety early warning system and early warning method thereof

Abstract

The invention discloses a reservoir dam flood discharge affected river reach safety early warning system and an early warning method thereof, and belongs to the technical field of dam flood discharge safety early warning of water conservancy and hydropower engineering. The system comprises an early-warning river reach and a plurality of unmanned aerial vehicle monitoring stations, a plurality of fixed targets are arranged on bank slopes on the two sides of the early-warning river reach, the fixed targets on the bank slopes on the two sides of the early-warning river reach are alternately arranged in a staggered mode, the unmanned aerial vehicle monitoring stations are arranged on the bank slopes of the early-warning river reach, and the unmanned aerial vehicle monitoring stations are provided with unmanned aerial vehicles. A high-definition camera, an RTK measuring instrument, a wireless transceiver module and a controller are carried on the unmanned aerial vehicle. The unmanned aerial vehicle can quickly capture a fixed target, and can quickly and accurately position a dangerous case position in combination with a high-precision RTK measuring instrument carried on the unmanned aerial vehicle, so that the position of a person involved in danger can be confirmed in time when the dangerous case occurs, the response timeliness of emergency rescue measures is improved, effective rescue measures are adopted in a targeted manner, and the emergency rescue efficiency is improved. And the risk of personnel casualties is reduced.

Description

Technical Field

[0001] The invention relates to a safety early warning system for a river section affected by flood discharge of a reservoir dam and an early warning method thereof, belonging to the technical field of safety early warning for flood discharge of a dam in a water conservancy and hydropower project. Background Art

[0002] During the operation of the reservoir dam, the reservoir water needs to be discharged according to the needs of flood control, power generation, dispatching, etc. During the reservoir discharge, the water level in the river section within a certain range downstream will rise rapidly and the flow rate will increase. Therefore, in order to ensure that the reservoir discharge does not cause loss of life and property. Existing laws and regulations require that the reservoir discharge should issue early warning notifications along the downstream river section.

[0003] The current specification "Technical Specifications for Hydropower Station Flood Discharge Warning Broadcasting System" requires that: in the river section 20 kilometers downstream from 300 meters downstream of the hydropower station dam axis, the local government organizes the hydropower station and the river management department to jointly determine the location and number of monitoring and early warning stations, and issue the first SMS warning one day before the reservoir discharges water, and issue the second SMS warning 30 minutes before the discharge and broadcast it, and broadcast the warning immediately in special circumstances. However, in actual implementation, the existing monitoring and early warning methods still have the following shortcomings:

[0004] 1. Since the flood discharge warning needs to cover the 20-kilometer river section downstream, the warning range is wide, the line is long, and the area is scattered. The existing single monitoring and warning stations are generally only arranged in typical areas, so the coverage is limited, there are blind spots for warning broadcasts, and the actual warning effect is limited;

[0005] 2. As the residents along the river are scattered and the activities of people are highly random, in order to ensure that all people in the river section affected by the flood discharge are evacuated to safe areas in a timely manner, in addition to the above-mentioned automatic warning broadcast, a large amount of manual inspection and notification work is still required, which actually requires a large workload and low efficiency;

[0006] 3. After a dangerous situation occurs, if the dangerous area is not within the coverage of the monitoring and early warning stations, it is difficult to promptly confirm the location, number and degree of danger of the people involved in the danger, resulting in untimely and inappropriate emergency rescue measures, resulting in a high risk of casualties;

[0007] 4. During the long-term operation of the existing single monitoring and early warning site devices, a large amount of human resources are required to carry out communication, power supply, lightning protection, anti-theft and maintenance work, which requires a high level of professionalism of the personnel and has high long-term investment costs.

[0008] The Chinese patent document with publication number CN206833664U discloses a flood discharge and drainage early warning monitoring system for a hydropower station, including an early warning host, a wireless signal transceiver of the early warning host and a plurality of unmanned aerial vehicles in communication connection, a controller is arranged on the unmanned aerial vehicle, and the controller is respectively connected with an audible and visual alarm, a video camera, an infrared thermal imager, a wireless alarm signal transmitting device, and a nighttime lighting lamp; the unmanned aerial vehicle and the brackets arranged on the upper ends of the street lamps on both sides of the downstream river channel of the hydropower station form a detachable connection structure. Before the hydropower station discharges flood water, the early warning host can control a plurality of unmanned aerial vehicles to patrol the downstream river channel, and the patrol coverage of the unmanned aerial vehicle is wide; when the unmanned aerial vehicle finds a ship or a person in the downstream river channel, the audible and visual alarm and the wireless alarm signal transmitting device can be used to respectively warn the pedestrian or the ship, and the early warning effect is good. The bracket can collect solar energy to charge the unmanned aerial vehicle, and the maintenance cost is low.

[0009] However, after a dangerous situation occurs, it is difficult for the system to promptly confirm the location of people in danger, resulting in untimely and inappropriate emergency rescue measures, and a high risk of casualties. Summary of the invention

[0010] In order to solve the above technical problems, the present invention provides a safety early warning system for river sections affected by flood discharge from a reservoir dam and an early warning method thereof.

[0011] The present invention is achieved through the following technical solutions:

[0012] A safety early warning system for a river section affected by flood discharge from a reservoir dam comprises an early warning river section and a plurality of unmanned aerial vehicle (UAV) monitoring stations, wherein a plurality of fixed targets are arranged on the bank slopes on both sides of the early warning river section, and the fixed targets on the bank slopes on both sides of the early warning river section are arranged alternately and staggered, and the plurality of UAV monitoring stations are arranged on the bank slopes of the early warning river section, and the UAV monitoring stations are equipped with UAVs, which are equipped with high-definition cameras, RTK measuring instruments, wireless transceiver modules and controllers, and the high-definition cameras, RTK measuring instruments and wireless transceiver modules are all electrically connected to the controller.

[0013] The fixed targets are GNSS reference stations. The fixed targets on the bank slopes on both sides of the warning river section are evenly spaced, and the spacing is L1.

[0014] The mileage along the warning river section is L, and N drone monitoring stations are evenly spaced on the bank slope of the warning river section, and the spacing between the N drone monitoring stations is L2, then N=L / L2.

[0015] Each of the drone monitoring stations is equipped with two drones.

[0016] The drone is equipped with an infrared thermal imaging night vision device and a loudspeaker, and both the infrared thermal imaging night vision device and the loudspeaker are electrically connected to the controller.

[0017] An early warning method for a safety early warning system of a river section affected by flood discharge from a reservoir dam, comprising the following steps:

[0018] Step 1: Use drone oblique photography to establish an overall 3D real-scene model of the warning river section and a real-scene 3D model of the surrounding L1×L1 range centered on each fixed target, and mark the spatial position coordinates of each drone monitoring station and fixed target in the overall 3D real-scene model;

[0019] Step 2: Enter the spatial position coordinates of each fixed target into the corresponding real-scene 3D model to ensure that after the drone recognizes the fixed target, the spatial position coordinates of the fixed target and the corresponding real-scene 3D model information are extracted and matched with the overall 3D real-scene model to achieve the restoration of the real-time scene at any time;

[0020] Step 3: Before and during the release of water from the reservoir dam, the two drones at each drone monitoring station will patrol along the river alternately. When one drone reaches the limit of its range, it will return to the drone monitoring station for charging, while the other drone will take off immediately to continue patrolling. This process will be carried out alternately to achieve all-weather patrols.

[0021] Step 4: Automatically analyze the data acquired by the drone. First, use the target detection algorithm to identify the person in the image and give the target's location and confidence score. Then analyze whether the heat emitted by the human body detected by the infrared thermal imaging night vision device exceeds the temperature threshold T set by the controller to distinguish between the human body and other heat sources.

[0022] Step 5. When personnel are identified as being active in the warning river section, the loudspeaker is automatically triggered to turn on, and calls are immediately made to drive the personnel away. At the same time, SLAM technology is used in combination with inertial measurement data to locate and track the movement paths of personnel in real time.

[0023] The specific method of step 2 comprises the following steps:

[0024] Step A, data processing: using GNSS to obtain the spatial position coordinate data of the fixed target, sorting out the model data of the real scene 3D model and the overall 3D real scene model, and performing format conversion and cleaning on the real scene 3D model to remove irrelevant information;

[0025] Step B, model fusion: extract feature points related to the fixed target from the real-scene 3D model, then use the feature matching algorithm to identify and match these feature points, and correspond the feature points in the real-scene 3D model to the spatial position coordinates of the fixed target. Next, use the 3D model fusion technology and the spatial position coordinates of the fixed target to spatially align the real-scene 3D model with the overall 3D real-scene model, so as to achieve seamless connection between the two models in space.

[0026] Step C, real-time scene restoration: match the real-time data of scene changes captured by the drone with the fused three-dimensional model in step B, use computer vision technology to update the scene, and continuously adjust the algorithm parameters to optimize the matching process, so as to achieve rapid extraction of fixed target positions and match them with the overall three-dimensional real-scene model, thereby restoring the real-time scene at any time.

[0027] The model data in step A includes but is not limited to terrain, buildings and vegetation;

[0028] The feature points in step B include but are not limited to color, texture and shape;

[0029] The three-dimensional model fusion technology in step B is point cloud registration or grid fusion.

[0030] In the step 4, motion is also detected by comparing the differences between consecutive frames to identify moving objects. The method for detecting motion is: M(x,y,t)=|I(x,y,t)-B(x,y)|, wherein M(x,y,t) is the motion intensity at position (x,y) at time t, I(x,y,t) is the pixel value of the current frame, and B(x,y) is the pixel value of the background model. The data of the high-definition camera and the infrared thermal imaging night vision device are then fused using the weighted average and feature-level fusion methods to comprehensively determine whether there is human activity in the warning river section.

[0031] In the step 5, the methods of steps 2 to 4 are also used to automatically match the on-site real-time scene with the overall three-dimensional real-scene model, and the Dijkstra or RRT path planning algorithm is used to calculate the best evacuation route from the current position to the designated safe area based on the terrain information in the fused three-dimensional model and the obstacle information in the real-time scene. Then, the loudspeaker is used to guide the on-site personnel to evacuate to the designated safe area, and the real-time on-site images are reported to the management center for decision-making; when the on-site personnel do not evacuate as required or people are in danger, the management center immediately notifies and organizes rescue personnel to rush to the scene for rescue.

[0032] The beneficial effects of the present invention are:

[0033] 1. Set up multiple drone monitoring stations at equal intervals on the banks of the warning river section. Before and during the water release from the reservoir dam, two drones from the drone monitoring station fly alternately along the river channel to achieve full coverage, all-weather patrols and real-time monitoring of the warning river section, thereby improving the timeliness of the warning.

[0034] 2. Multiple fixed targets are set up at equal intervals and alternately on the slopes on both sides of the warning river section to ensure that the drone can quickly capture the fixed targets. The three-dimensional real-scene model established by the drone's oblique photography, combined with the high-precision RTK measuring instrument on the drone, can achieve rapid and accurate positioning of the dangerous situation, thereby improving the accuracy of positioning. When a dangerous situation occurs, the location of the people in danger can be confirmed in time to improve the timeliness of the emergency rescue measures, and effective rescue measures can be taken in a targeted manner to reduce the risk of casualties.

[0035] 3. Utilize target detection algorithms and infrared thermal imaging technology, combined with data collected by high-definition cameras, automatically analyze data obtained by drones, effectively identify and track human activities in the warning river section, and improve the accuracy of identifying human activities and the speed of response.

[0036] 4. When human activities are detected in the warning river section, the loudspeaker immediately shouts to drive away the on-site personnel, and combines SLAM technology and path planning algorithms to calculate the best evacuation route to guide on-site personnel to evacuate safely. At the same time, the management center can organize rescue in time, which improves the intelligence level of the early warning system and ensures the safety of on-site personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a structural schematic diagram of the present invention.

[0038] In the picture: 1-UAV monitoring station, 3-fixed target, 4-early warning river section. DETAILED DESCRIPTION

[0039] The technical solution of the present invention is further described below, but the scope of protection claimed is not limited to the description.

[0040] like Figure 1 As shown, a safety early warning system for a river section affected by flood discharge of a reservoir dam described in the present invention comprises an early warning river section 4 and a plurality of unmanned aerial vehicle monitoring stations 1, a plurality of fixed targets 3 are arranged on the bank slopes on both sides of the early warning river section 4, and the fixed targets 3 on the bank slopes on both sides of the early warning river section 4 are arranged alternately and staggered, and a plurality of the unmanned aerial vehicle monitoring stations 1 are arranged on the bank slopes of the early warning river section 4, and the unmanned aerial vehicle monitoring station 1 is equipped with an unmanned aerial vehicle, and the unmanned aerial vehicle is equipped with a high-definition camera, an RTK measuring instrument, a wireless transceiver module and a controller, and the high-definition camera, the RTK measuring instrument and the wireless transceiver module are all electrically connected to the controller. When in use, the early warning river section 4 is the river section affected by flood discharge of the reservoir dam, and the unmanned aerial vehicle monitoring station 1 is preferably set in a location with open terrain, reliable communication and convenient power supply.

[0041] The fixed targets 3 are GNSS reference stations, and the fixed targets 3 on the bank slopes on both sides of the warning river section 4 are arranged at equal intervals, and the arrangement interval is L1. When in use, L1 = 500m, that is, multiple fixed targets 3 are alternately arranged on the left and right banks of the warning river section 4 at an interval of 500m, ensuring that when the drone flies to any position, it can quickly capture the three fixed targets 3 closest to it, so as to accurately and quickly locate the position of the drone.

[0042] The mileage along the warning river section 4 is L, and N drone monitoring stations 1 are evenly spaced on the bank slope of the warning river section 4, and the spacing between the N drone monitoring stations 1 is L2, then N = L / L2. When in use, L2 = 5km, to ensure that a single drone monitoring station 1 can cover the upstream and downstream sections of 2.5km each centered on it, and the total number of drone monitoring stations 1 arranged along the warning river section 4 is N = L / 5.

[0043] Each of the drone monitoring stations 1 is equipped with two drones.

[0044] The drone is equipped with an infrared thermal imaging night vision device and a loudspeaker, and both are electrically connected to the controller. When in use, the RTK measuring instrument is used to dynamically obtain the real-time spatial position coordinates of the drone during flight; the drone is equipped with a high-definition camera and an infrared thermal imaging night vision device to achieve daytime data collection; the drone is equipped with a loudspeaker for on-site broadcasting and shouting.

[0045] An early warning method for a safety early warning system of a river section affected by flood discharge from a reservoir dam, comprising the following steps:

[0046] Step 1: Use drone oblique photography to establish an overall three-dimensional real-scene model of the warning river section 4 and a real-scene three-dimensional model within the surrounding L1×L1 range centered on each fixed target 3, and mark the spatial position coordinates of each drone monitoring station 1 and fixed target 3 in the overall three-dimensional real-scene model.

[0047] Step 2: Enter the spatial position coordinates of each fixed target 3 into the corresponding real-scene three-dimensional model to ensure that after the drone recognizes the fixed target 3, the spatial position coordinates of the fixed target 3 and the corresponding real-scene three-dimensional model information are extracted and matched with the overall three-dimensional real-scene model to achieve the restoration of the real-time scene at any time;

[0048] Step 3: Before and during the release of water from the reservoir dam, the two drones at each drone monitoring station 1 will patrol along the river alternately. When one drone reaches the limit of its range, it will return to drone monitoring station 1 for charging, while the other drone will take off immediately to continue patrolling. This process will be repeated alternately to achieve all-weather patrols. Each drone will autonomously fly back and forth within the 5km river range it covers.

[0049] Step 4: Automatically analyze the data acquired by the drone. First, use the target detection algorithm to identify the person in the image and give the target's location and confidence score. Then analyze whether the heat emitted by the human body detected by the infrared thermal imaging night vision device exceeds the temperature threshold T set by the controller to distinguish between the human body and other heat sources.

[0050] Step 5. When personnel are identified to be active in the warning river section 4, the loudspeaker is automatically triggered to turn on, and calls are immediately made to drive the personnel away. At the same time, SLAM technology is used in combination with inertial measurement data to locate and track the movement paths of the personnel in real time.

[0051] The specific method of step 2 comprises the following steps:

[0052] Step A, data processing: Use GNSS to obtain the spatial position coordinate data of the fixed target 3, organize the model data of the real scene 3D model and the overall 3D real scene model, and convert and clean the real scene 3D model to remove irrelevant information. Irrelevant information includes noise, redundant details, etc.

[0053] Step B, model fusion: extract feature points related to the fixed target 3 from the real-life 3D model, then use a feature matching algorithm to identify and match these feature points, and correspond the feature points in the real-life 3D model to the spatial position coordinates of the fixed target 3. Next, use the 3D model fusion technology and utilize the spatial position coordinates of the fixed target 3 to spatially align the real-life 3D model with the overall 3D real-life model to achieve seamless connection between the two models in space.

[0054] Step C, real-time scene restoration: match the real-time data of scene changes captured by the drone with the fused three-dimensional model in step B, use computer vision technology to update the scene, and continuously adjust the algorithm parameters to optimize the matching process, so as to achieve rapid extraction of the position of the fixed target 3 and match it with the overall three-dimensional real scene model, thereby restoring the real-time scene at any time.

[0055] The model data in step A includes but is not limited to terrain, buildings and vegetation;

[0056] The feature points in step B include but are not limited to color, texture and shape;

[0057] The three-dimensional model fusion technology in step B is point cloud registration or grid fusion.

[0058] In the step 4, the motion is detected by comparing the difference between consecutive frames to identify moving objects. The method for detecting motion is: M(x, y, t) = |I(x, y, t)-B(x, y)|, where M(x, y, t) is the motion intensity at time t at position (x, y), I(x, y, t) is the pixel value of the current frame, and B(x, y) is the pixel value of the background model. Then, the data of the high-definition camera and the infrared thermal imaging night vision device are fused using the weighted average and feature-level fusion methods to comprehensively determine whether there is human activity in the warning river section 4. The method for detecting motion can improve the accuracy of identifying human activities.

[0059] In the step 5, the method of step 2 to step 4 is also used to automatically match the real-time scene on site with the overall three-dimensional real-scene model, and the Dijkstra or RRT path planning algorithm is used to calculate the best evacuation route from the current position to the designated safe area based on the terrain information in the fused three-dimensional model and the obstacle information in the real-time scene, and then the on-site personnel are guided to evacuate to the designated safe area through a loudspeaker, and the real-time scene on site is reported to the management center for decision-making; when the on-site personnel do not evacuate as required or there is a situation where people are in danger, the management center immediately notifies and organizes rescue personnel to rush to the scene for rescue. When in use, the drone is wirelessly connected to the command center console of the management center through the wireless transceiver module on it.

Claims

1. A safety early warning system for river sections affected by flood discharge from a reservoir dam, characterized by: The invention comprises an early warning river section (4) and a plurality of unmanned aerial vehicle monitoring stations (1), wherein a plurality of fixed targets (3) are arranged on the bank slopes on both sides of the early warning river section (4), and the fixed targets (3) on the bank slopes on both sides of the early warning river section (4) are arranged alternately and staggered, and the plurality of unmanned aerial vehicle monitoring stations (1) are arranged on the bank slopes of the early warning river section (4), and the unmanned aerial vehicle monitoring stations (1) are equipped with unmanned aerial vehicles, and the unmanned aerial vehicles are equipped with high-definition cameras, RTK measuring instruments, wireless transceiver modules and controllers, and the high-definition cameras, RTK measuring instruments and wireless transceiver modules are all electrically connected to the controller.

2. The safety early warning system for river sections affected by flood discharge from reservoir dams as claimed in claim 1, characterized in that: The fixed targets (3) are GNSS reference stations. The fixed targets (3) on the bank slopes on both sides of the warning river section (4) are arranged at equal intervals, and the arrangement interval is L1.

3. The safety early warning system for river sections affected by flood discharge from reservoir dams as claimed in claim 1, characterized in that: The mileage along the warning river section (4) is L, and N drone monitoring stations (1) are evenly spaced on the bank slope of the warning river section (4), and the spacing between the N drone monitoring stations (1) is L2, then N=L / L2.

4. The safety early warning system for river sections affected by flood discharge from a reservoir dam as claimed in claim 1 or 3, characterized in that: Each of the drone monitoring stations (1) is equipped with two drones.

5. The safety early warning system for river sections affected by flood discharge from reservoir dams as claimed in claim 1, characterized in that: The drone is equipped with an infrared thermal imaging night vision device and a loudspeaker, and both the infrared thermal imaging night vision device and the loudspeaker are electrically connected to the controller.

6. An early warning method for a safety early warning system for a river section affected by flood discharge from a reservoir dam as claimed in any one of claims 1 to 5, characterized in that: The following steps are involved: Step 1: Use drone oblique photography to establish an overall three-dimensional real-scene model of the warning river section (4) and a real-scene three-dimensional model of the surrounding L1×L1 range centered on each fixed target (3), and mark the spatial position coordinates of each drone monitoring station (1) and the fixed target (3) in the overall three-dimensional real-scene model; Step 2: entering the spatial position coordinates of each fixed target (3) into the corresponding real-scene three-dimensional model, ensuring that after the drone recognizes the fixed target (3), the spatial position coordinates of the fixed target (3) and the corresponding real-scene three-dimensional model information are extracted and matched with the overall three-dimensional real-scene model to achieve restoration of the real-time scene at any time; Step 3: Before and during the release of water from the reservoir dam, the two drones at each drone monitoring station (1) conduct patrols along the river alternately, i.e., after one drone reaches the end of its range, it returns to the drone monitoring station (1) for charging, while the other drone immediately takes off and continues patrolling, and the patrols are carried out alternately in this way, thus achieving all-weather patrols; Step 4: Automatically analyze the data acquired by the drone. First, use the target detection algorithm to identify the person in the image and give the target's location and confidence score. Then analyze whether the heat emitted by the human body detected by the infrared thermal imaging night vision device exceeds the temperature threshold T set by the controller to distinguish between the human body and other heat sources. Step 5: When it is identified that there are people in the warning river section (4), the loudspeaker is automatically triggered to turn on and immediately shout to drive the people away. At the same time, SLAM technology is used in combination with inertial measurement data to locate and track the movement path of the people in real time.

7. The early warning method of the safety early warning system of the river section affected by the flood discharge of the reservoir dam as claimed in claim 6, characterized in that: The specific method of step 2 comprises the following steps: Step A, data processing: using GNSS to obtain spatial position coordinate data of the fixed target (3), collating model data of the real scene three-dimensional model and the overall three-dimensional real scene model, and performing format conversion and cleaning on the real scene three-dimensional model to remove irrelevant information; Step B, model fusion: extract feature points related to the fixed target (3) from the real-scene three-dimensional model, then use a feature matching algorithm to identify and match these feature points, and correspond the feature points in the real-scene three-dimensional model to the spatial position coordinates of the fixed target (3). Next, use the three-dimensional model fusion technology and the spatial position coordinates of the fixed target (3) to spatially align the real-scene three-dimensional model with the overall three-dimensional real-scene model, so as to achieve seamless connection between the two models in space. Step C, real-time scene restoration: match the real-time data of scene changes captured by the drone with the three-dimensional model fused in step B, use computer vision technology to update the scene, and continuously adjust the algorithm parameters to optimize the matching process, so as to achieve rapid extraction of the position of the fixed target (3) and match it with the overall three-dimensional real scene model, thereby restoring the real-time scene at any time.

8. The early warning method of the safety early warning system of the river section affected by the flood discharge of the reservoir dam as claimed in claim 7, characterized in that: The model data in step A includes but is not limited to terrain, buildings and vegetation; The feature points in step B include but are not limited to color, texture and shape; The three-dimensional model fusion technology in step B is point cloud registration or grid fusion.

9. The early warning method of the reservoir dam flood discharge affected river section safety early warning system as claimed in claim 6, characterized in that: In step 4, motion is detected by comparing the differences between consecutive frames to identify moving objects. The method for detecting motion is: M(x,y,t)=|I(x,y,t)-B(x,y)|, where M(x,y,t) is the motion intensity at position (x,y) at time t, I(x,y,t) is the pixel value of the current frame, and B(x,y) is the pixel value of the background model. Then, the data of the high-definition camera and the infrared thermal imaging night vision device are fused using the weighted average and feature-level fusion methods to comprehensively determine whether there is any human activity in the warning river section (4).

10. The early warning method of the safety early warning system of the river section affected by the flood discharge of the reservoir dam as claimed in claim 6, characterized in that: In the step 5, the methods of steps 2 to 4 are also used to automatically match the on-site real-time scene with the overall three-dimensional real-scene model, and the Dijkstra or RRT path planning algorithm is used to calculate the best evacuation route from the current position to the designated safe area based on the terrain information in the fused three-dimensional model and the obstacle information in the real-time scene. Then, the loudspeaker is used to guide the on-site personnel to evacuate to the designated safe area, and the real-time on-site images are reported to the management center for decision-making; when the on-site personnel do not evacuate as required or people are in danger, the management center immediately notifies and organizes rescue personnel to rush to the scene for rescue.

Images