Disaster identification device based on unmanned aerial vehicle

Abstract

The invention discloses a disaster recognition device based on an unmanned aerial vehicle, which comprises an aerial vehicle, a detection box, a tail end box and a signal terminal, and is characterized in that the detection box is deployed below the aerial vehicle, and a camera is arranged on the detection box; the tail end box is provided with a plurality of sensing units, the tail end box is connected to the detection box through a cable, and the tail end box is suspended below the detection box through the cable; the aircraft and the detection box are both in communication connection with the signal terminal and are used for receiving data of the detection box and interacting with a user. The method has the advantages that more accurate fire information of a fire area can be acquired, and adverse effects caused by smoke and high temperature on a fire scene are overcome.

Description

Technical Field

[0001] The present application relates to the field of firefighting and rescue technology, and in particular to a disaster identification device based on an unmanned aerial vehicle. Background Art

[0002] Industrial sites such as chemical parks have a high risk of fire due to their special production environment and flammable and explosive materials. Traditional fire monitoring methods, such as fixed smoke detectors and manual inspections, have defects such as limited monitoring range, slow response speed, and inability to cover large areas. In addition, once a fire occurs, due to the lack of timely and accurate fire information, firefighters often find it difficult to quickly take effective fire-fighting measures, resulting in the spread of the fire and causing greater losses. At present, drones rely solely on cameras for visual recognition in terms of fire identification. However, this method has recognition errors because drones cannot get close to the fire source for more accurate monitoring. In addition, current drone gas detection is often hard-wired to drones, requiring drones to get close to the fire source to identify smoke components, which will lead to fire and collision risks for drones and the camera cannot see the scene clearly due to smoke. Therefore, for fire monitoring in industrial sites such as chemical parks, more advanced, accurate, and efficient monitoring methods are needed to improve fire identification, early warning, and emergency response capabilities and reduce fire losses. Summary of the invention

[0003] The present application provides a disaster identification device based on an unmanned aerial vehicle, which has the advantage of being conducive to obtaining more accurate fire information in the fire area and overcoming the adverse effects of smoke and high temperature at the fire scene.

[0004] The above-mentioned object of the present application is achieved through the following technical solutions: a disaster identification device based on an unmanned aerial vehicle, comprising: Aircraft; A detection box is deployed under the aircraft and is equipped with a camera; At least one terminal box, the terminal box is provided with a plurality of sensor units, the terminal box is connected to the detection box through cables, and the terminal box is suspended below the detection box through the cables; And a signal terminal, the aircraft and the detection box are both connected to the signal terminal for communication, and is used to receive data from the detection box and interact with the user.

[0005] Furthermore, the detection box is provided with a first circuit board and a power module, the terminal box is connected to the first circuit board via a cable, and the camera is connected to the first circuit board; The detection box is provided with a cable winder for winding the cable, the cable winder is connected to a motor for driving the cable winder to rotate, and the motor is connected to and controlled by the first circuit board.

[0006] Furthermore, a through hole for the cable to pass through is provided at the bottom of the detection box, and an elastic member is provided at the top of the terminal box, and the elastic member is used to contact the bottom of the detection box and provide thrust for the terminal box.

[0007] Furthermore, an electromagnet is provided at the bottom of the detection box, and a suction member used in conjunction with the electromagnet is provided at the top of the terminal box.

[0008] Furthermore, the terminal boxes are provided with four groups, and the detection boxes are provided with four groups of winders and four groups of electromagnets corresponding to the terminal boxes one by one; the lengths of the cables connected to the four groups of terminal boxes increase sequentially.

[0009] Furthermore, a second circuit board is provided in the terminal box, the sensor unit is connected to the second circuit board, the terminal box has a ventilation hole running through it from top to bottom, the second circuit board is arranged on the side of the ventilation hole, and the sensor unit is located in the ventilation hole.

[0010] Furthermore, the sensing unit is a carbon monoxide sensor, a carbon dioxide sensor, a hydrocarbon sensor, a smoke sensor and a nitrogen oxide sensor.

[0011] Furthermore, the signal terminal includes a fire detection unit, a path planning unit and a sensor data analysis unit; wherein, The fire detection unit deploys an image recognition algorithm to analyze the image data obtained by the camera, identify the flames and smoke in the image and mark them; The path planning unit is used to plan the flight path of the aircraft so that the aircraft moves above the flame or smoke; The sensor signal analysis unit is used to analyze the detection data of the terminal box, and the steps are as follows: S1: Acquire the data detected by the sensor unit and determine the location of the sensor unit according to the position and altitude of the aircraft; S2: comparing the detection data of the sensor unit with a first threshold value, and determining that a fire exists at the target location when the detection data exceeds the first threshold value; S3: acquiring detection data of four groups of terminal boxes when the aircraft is at the same position, acquiring the corresponding relationship between the detection data and the altitude, calculating the change rate of the detection data in altitude according to the corresponding relationship between the detection data and the altitude, and determining the fire level according to the change rate; S4: The fire information obtained in step S2, the fire level obtained in step S3, and the target area image obtained by the camera are integrated and output.

[0012] Furthermore, in step S3, the fire level is divided into four levels of A, B, C, and D from low to high according to the rate of change in height of the terminal box detection data.

[0013] In summary, the beneficial effects of this application are: 1. This application collects fire information by carrying a lowerable terminal box on an aircraft, which is helpful to overcome the influence of smoke and high temperature, obtain more accurate fire information, and reduce the adverse effects of high temperature, smoke, etc. on the aircraft; 2. This application collects information at different heights by carrying terminal boxes at different heights. According to the detection information at different heights, the severity of the fire in the target area can be obtained, which is conducive to obtaining a more accurate fire map and fire spread to facilitate rescue. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the overall structure of an embodiment of the present application; Figure 2 It is a schematic diagram of the structure of the terminal box of an embodiment of the present application; Figure 3 It is a top view schematic diagram of an end box according to an embodiment of the present application.

[0015] In the figure, 100, aircraft; 200, detection box; 201, winder; 202, electromagnet; 300, terminal box; 301, sensor unit; 302, elastic member; 303, suction member; 400, cable; 500, signal terminal. DETAILED DESCRIPTION

[0016] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0017] Embodiment: The above-mentioned object of the present application is achieved through the following technical solution: a disaster identification device based on an unmanned aerial vehicle, referring to Figure 1 , including an aircraft, a detection box, a terminal box and a signal terminal.

[0018] The aircraft is a drone; the detection box is deployed under the aircraft. In this embodiment, the detection box and the aircraft are detachably connected, and are independently powered and controlled, so the detection box can be deployed on different aircraft.

[0019] The detection box is provided with a camera for collecting image data of the target area.

[0020] Four groups of terminal boxes are provided, and the terminal boxes are provided with a plurality of sensor units. The terminal boxes are connected to the detection box through cables, and the terminal boxes are suspended below the detection box through the cables.

[0021] The detection box is provided with a first circuit board and a power module, the terminal box is connected to the first circuit board through a cable, and the camera is connected to the first circuit board; the detection box is provided with a winder for winding the cable, the winder is connected to a motor for driving the winder to rotate, and the motor is connected to the first circuit board and is controlled. There are four groups of winders, which are used in conjunction with four groups of terminal boxes respectively.

[0022] In this embodiment, the cables connected to the four groups of terminal boxes are of different lengths, and the lengths of the cables connected to the four groups of terminal boxes increase in sequence. Typically, the lengths of the cables connected to the four groups of terminal boxes are 20m, 15m, 10m, and 5m. The winder is a turntable winder, which is driven and controlled by a motor. The cable uses high-temperature fluoroplastic insulated wire, has good flexibility and high temperature resistance, and can work for a long time in a high temperature environment. In this embodiment, a conductive slip ring is also provided at the end of the cable to cooperate with the winder.

[0023] The bottom of the detection box is provided with a through hole for the cable to pass through, and the top of the terminal box is provided with an elastic member, which is used to contact the bottom of the detection box and provide thrust for the terminal box. In this embodiment, the elastic member is a spring.

[0024] The bottom of the detection box is provided with an electromagnet, and the top of the terminal box is provided with an attraction member (such as Figure 3 ), the electromagnet is connected to the first circuit board and controlled.

[0025] refer to Figure 2-3 The terminal box is provided with a second circuit board, the sensor unit is connected to the second circuit board, the terminal box has a ventilation hole running through it from top to bottom, the second circuit board is arranged on the side of the ventilation hole, and the sensor unit is located in the ventilation hole. The terminal box is provided with a partition, both sides of the partition are ventilation holes, the second circuit board is installed on the side of the partition, and a number of sensor units are arranged on the second circuit board. The second circuit board collects data from the sensor unit and transmits it to the detection box through cables.

[0026] The sensing units are carbon monoxide sensors, carbon dioxide sensors, hydrocarbon sensors, smoke sensors and nitrogen oxide sensors.

[0027] The aircraft and the detection box are both connected to the signal terminal for communication, and the signal terminal is used to receive data from the detection box and interact with the user. The signal terminal is also used to send control instructions to the detection box and the aircraft to control the flight and hovering of the aircraft and control the motor and electromagnet in the detection box.

[0028] The signal terminal includes a fire detection unit, a path planning unit and a sensor data analysis unit; wherein, The fire detection unit deploys an image recognition algorithm to analyze the image data obtained by the camera, identify the flame and smoke in the image and mark them; specifically, in this embodiment, the fire detection unit deploys a trained improved YOLOv3 algorithm to identify flame and smoke. The improvement of the YOLOv3 algorithm is to add a yolo layer in the feature interaction stage to enhance the feature fusion.

[0029] The flame identification process is as follows: The original images acquired by the drone camera are preprocessed by resizing and other operations to make them meet the requirements of network input. For example, they are uniformly adjusted to a specific size (such as common sizes such as 416×416 pixels). At the same time, the images are normalized and the pixel values ​​are mapped to a suitable range (such as between 0 and 1) to facilitate subsequent calculations.

[0030] Darknet-53 is used as the basic feature extraction network structure (this is the core feature extraction part of YOLOv3). Through a series of convolutional layers (Convolutional Layer) and pooling layers (YOLOv3 uses convolutional layers with a step size of 2 to replace some pooling layers to achieve downsampling, so as to better retain feature information), feature maps of different levels are gradually extracted from the input image.

[0031] In the improved version of YOLOv3, the yolo layer is added in the feature interaction stage to enhance the feature fusion. Feature maps at different levels are fused, for example, by upsampling the low-level feature map and then concatenating it with the high-level feature map, so that the network can learn both the detailed information of the image (from the low-level features) and the abstract semantic information (from the high-level features). The fused feature map is then used for multi-scale prediction. YOLOv3 usually performs target detection prediction on feature maps of 3 different scales, and divides them into grids of different sizes. Each grid is responsible for predicting the target-related information that falls within the grid.

[0032] For each grid, multiple bounding boxes and their corresponding category probabilities are predicted. The predicted output tensor contains the coordinate information of the bounding box (such as the center coordinates, width, and height. The coordinates here are relative coordinates to the grid. After a certain conversion, the absolute coordinates in the original image can be obtained), confidence score (indicates the probability that the bounding box contains the target and the prediction is accurate) and category probability (for example, for the two categories of fire and non-fire, it is the corresponding probability value of fire).

[0033] The confidence calculation takes into account the probability of the target's existence and the accuracy of the prediction box, and calculates a value between 0 and 1 through the weights and features learned by the network. The category probability is converted into a probability value by using activation functions such as Softmax (for multi-category situations) or Sigmoid function (for binary classification such as fire and non-fire situations) to indicate the possibility of belonging to each category.

[0034] Through post-processing methods such as non-maximum suppression (NMS), duplicate and redundant predicted bounding boxes are removed to screen out the bounding boxes that are most likely to represent the fire target. The basic idea is to retain bounding boxes with high confidence and IoU with other boxes less than a certain threshold as the final detection results based on indicators such as confidence and intersection over union (IoU, which calculates the ratio of the overlapping area of ​​two bounding boxes to the area of ​​the union).

[0035] When the visual recognition algorithm detects the position of the fire in the image (usually expressed in coordinates in the image coordinate system, such as the upper left corner of the image as the origin, the horizontal right as the positive direction of the x-axis, and the vertical downward as the positive direction of the y-axis), it is necessary to combine the current attitude of the drone (such as heading angle, roll angle, pitch angle) and the installation parameters of the camera (including the camera's intrinsic parameter matrix, involving parameters such as focal length, and the extrinsic parameter matrix reflects the position and attitude relationship of the camera relative to the drone body) to convert the coordinates of the fire point in the image to the coordinate system relative to the drone body. If the fire point is to be located in a geographic coordinate system (such as longitude and latitude coordinates), it is also necessary to combine the drone's own positioning system (such as GPS, etc.) to obtain the drone's current geographic coordinates and heading information, and calculate the geographic coordinates of the fire point through a coordinate transformation algorithm (involving the calculation of rotation matrices, etc., and converting the coordinates in the body coordinate system to the geographic coordinate system based on the drone's attitude angle).

[0036] The path planning unit is used to plan the flight path of the aircraft so that the aircraft moves above the flame or smoke. The path planning unit uses a path planning algorithm, such as the A algorithm, the Dijkstra algorithm, and other graph search-based algorithms to perform path planning. First, the flight area is discretized into a grid map (or a topological map of the constructed environment, etc.), the area where the obstacle is located is marked as an inaccessible area, the starting position of the drone and the location of the fire point are used as nodes, and the weight of the edge can be set to the distance between the two points or the flight cost (such as considering factors that affect flight energy consumption such as wind direction and airflow).

[0037] According to the planned path direction (such as the direction angle calculated above), the UAV's flight control system adjusts its attitude. Common attitude control algorithms based on PID (proportional-integral-differential) controllers are used to control the roll angle, pitch angle, and yaw angle of the UAV.

[0038] The sensor signal analysis unit is used to analyze the detection data of the terminal box, and the steps are as follows: S1: Acquire the data detected by the sensor unit and determine the location of the sensor unit according to the position and altitude of the aircraft; S2: comparing the detection data of the sensor unit with a first threshold value, and determining that a fire exists at the target location when the detection data exceeds the first threshold value; S3: When a fire is detected at the target location, the detection data of the four groups of terminal boxes when the aircraft is at the same location are obtained, the corresponding relationship between the detection data and the altitude is obtained, the change rate of the detection data in altitude is calculated according to the corresponding relationship between the detection data and the altitude, and the fire level is determined according to the change rate; S4: The fire information obtained in step S2, the fire level obtained in step S3, and the target area image obtained by the camera are integrated and output.

[0039] In step S3, the fire level is divided into four levels: A, B, C, and D according to the change rate of the terminal box detection data in height from low to high. The corresponding relationship is shown in the following table. In the table, R A , R B , R C is the preset threshold value.

[0040] Fire level The rate of change of the terminal box detection data in height A <![CDATA[0-R A ]]> B <![CDATA[R A -R B ]]> C <![CDATA[R B -R C ]]> D <![CDATA[>R C ]]> If a fire in a certain area is serious, the carbon monoxide, carbon dioxide, hydrocarbons, smoke sensors and nitrogen oxides produced by it will affect a larger area and height. Therefore, according to the terminal boxes located at different heights, the change rate of the above indicators in height can be obtained, and the fire level can be obtained, among which level A is the most serious and level D is the least serious. When fused with the regional image, different fire levels are marked with different colors.

[0041] The workflow of this embodiment is as follows: first, the aircraft is controlled to fly over an area where a fire has occurred or may occur, the camera collects images of the target area, the fire detection unit detects flames or smoke in the image, and sends a signal when flames or smoke are detected. At the same time, the path planning unit plans the path of the drone and controls the drone to fly above the flames or smoke. The signal terminal sends a control signal to the detection box to control the electromagnet to cut off the power and the motor to rotate, so that different terminal boxes are located at different heights below the terminal box to collect relevant signals of the target area. After the detection signal of the sensor unit is transmitted to the signal terminal, the sensor signal analysis unit analyzes and obtains the fire information and fire level information of a smaller area. As the aircraft moves, the fire information and fire level information of the entire target area are obtained to guide rescue.

[0042] The above is only a preferred implementation of the present application. It should be pointed out that a person skilled in the art can make several modifications and improvements without departing from the inventive concept of the present application, and these all fall within the scope of protection of the present application.

Claims

1. A disaster identification device based on an unmanned aerial vehicle, characterized in that: include: Aircraft; A detection box is deployed under the aircraft and is equipped with a camera; At least one terminal box, the terminal box is provided with a plurality of sensor units, the terminal box is connected to the detection box through cables, and the terminal box is suspended below the detection box through the cables; And a signal terminal, the aircraft and the detection box are both connected to the signal terminal for communication, and is used to receive data from the detection box and interact with the user.

2. The disaster identification device based on unmanned aerial vehicle according to claim 1, characterized in that: The detection box is provided with a first circuit board and a power module, the terminal box is connected to the first circuit board via a cable, and the camera is connected to the first circuit board; The detection box is provided with a cable winder for winding the cable, the cable winder is connected to a motor for driving the cable winder to rotate, and the motor is connected to and controlled by the first circuit board.

3. The disaster identification device based on unmanned aerial vehicle according to claim 2 is characterized in that: The bottom of the detection box is provided with a through hole for the cable to pass through, and the top of the terminal box is provided with an elastic member, which is used to contact the bottom of the detection box and provide a thrust for the terminal box.

4. The disaster identification device based on unmanned aerial vehicle according to claim 3 is characterized in that: The bottom of the detection box is provided with an electromagnet, and the top of the terminal box is provided with an attraction piece used in conjunction with the electromagnet.

5. The disaster identification device based on unmanned aerial vehicle according to claim 3 is characterized in that: The terminal boxes are provided with four groups, and the detection boxes are provided with four groups of winders and four groups of electromagnets corresponding to the terminal boxes one by one; the lengths of the cables connected to the four groups of terminal boxes increase sequentially.

6. The disaster identification device based on unmanned aerial vehicle according to claim 5, characterized in that: A second circuit board is arranged in the terminal box, the sensor unit is connected to the second circuit board, a ventilation hole is provided in the terminal box which passes through the terminal box from top to bottom, the second circuit board is arranged on the side of the ventilation hole, and the sensor unit is located in the ventilation hole.

7. The disaster identification device based on unmanned aerial vehicle according to claim 6 is characterized in that: The sensing units are carbon monoxide sensors, carbon dioxide sensors, hydrocarbon sensors, smoke sensors and nitrogen oxide sensors.

8. The disaster identification device based on unmanned aerial vehicle according to claim 7, characterized in that: The signal terminal includes a fire detection unit, a path planning unit and a sensor data analysis unit; wherein, The fire detection unit deploys an image recognition algorithm to analyze the image data obtained by the camera, identify the flames and smoke in the image and mark them; The path planning unit is used to plan the flight path of the aircraft so that the aircraft moves above the flame or smoke; The sensor signal analysis unit is used to analyze the detection data of the terminal box, and the steps are as follows: S1: Acquire the data detected by the sensor unit and determine the location of the sensor unit according to the position and altitude of the aircraft; S2: comparing the detection data of the sensor unit with a first threshold value, and determining that a fire exists at the target location when the detection data exceeds the first threshold value; S3: acquiring detection data of four groups of terminal boxes when the aircraft is at the same position, acquiring the corresponding relationship between the detection data and the altitude, calculating the change rate of the detection data in altitude according to the corresponding relationship between the detection data and the altitude, and determining the fire level according to the change rate; S4: The fire information obtained in step S2, the fire level obtained in step S3, and the target area image obtained by the camera are integrated and output.

9. The disaster identification device based on unmanned aerial vehicle according to claim 8, characterized in that: In step S3, the fire level is divided into four levels: A, B, C, and D from low to high according to the rate of change in height of the terminal box detection data.

Images