Inspection unmanned aerial vehicle for forest farm fire and operation method of inspection unmanned aerial vehicle

Abstract

The invention discloses a polling unmanned aerial vehicle for forest fire and an operation method thereof.The polling unmanned aerial vehicle is composed of four-axis drive and can stably fly in various environmental conditions, a dual-spectrum camera is arranged at the lower end of the polling unmanned aerial vehicle, and a protection mechanism is arranged on the surface of the polling unmanned aerial vehicle; the protection mechanism reduces the influence of environmental factors on the equipment performance as much as possible to prolong the service life of the equipment, a processor is arranged in the inspection unmanned aerial vehicle and is responsible for receiving and processing data of the dual-spectrum camera, and the inspection unmanned aerial vehicle is made of a flame-retardant material, so that the inspection unmanned aerial vehicle can keep certain working stability in an extremely high temperature environment.

Description

Technical Field

[0001] The present invention relates to unmanned aerial vehicle (UAV) technology and fire early warning and control technology, and in particular to a design of an inspection UAV for forest fires. Background Art

[0002] With the rapid development of social economy and moderate population growth in recent years, the protection of forest resources has become increasingly important, among which the prevention and control of forest fires is particularly critical. Traditional forest fire prevention and control methods, such as manual patrols, satellite remote sensing and fixed fire sensors, are often unable to meet the needs of timely detection and extinguishing of fire sources due to human resource limitations, long response time and small coverage area.

[0003] In recent years, drones have played an important role in forest fire prevention and control due to their flexibility and real-time performance. They can quickly reach the disaster site for inspection and reporting, and the cameras they carry can transmit images of the fire source in real time.

[0004] However, most drones on the market currently rely on manual fire source identification. Since the accuracy of fire source identification by the human eye is affected by many factors, such as the size and brightness of the fire source, this method has the risk of misidentification and missed identification. At the same time, the speed of manual fire source identification is relatively slow and often cannot meet the needs of rapid fire fighting.

[0005] On the other hand, due to the complexity and extreme nature of the forest environment, drones face many challenges when performing fire prevention and control tasks. For example, drones may encounter strong winds, extremely high temperatures, or smoke from fire sources, all of which may cause problems for the drone's stable flight. Due to the uncertainty and variability of fire sources, drones also have great difficulties in positioning and navigating in complex environments. Summary of the invention

[0006] The purpose of the present invention is to address the problem that existing UAVs in the prior art mainly rely on manual operation to identify fire sources and work abnormally or fail in complex and extreme environments. The purpose of the present invention is to provide a quadcopter UAV design that is optimized for fire inspection, can autonomously identify and locate fire sources, and perform reliable autonomous flight and decision-making in various environments.

[0007] Technical solution:

[0008] A patrol drone for forest fires comprises a main control module. A wireless communication module is disposed above the main control module and is connected to a flight module below the main control module. The flight module is connected to a landing module below the main control module, and the landing module is connected to an identification module below the main control module.

[0009] As the preferred technical solution of the present application, the main control module of the inspection UAV includes a main control support plate, an embedded processor is placed above the main control support plate, the lower end of the main control support plate is connected to the upper plate of the UAV through four upper pads, the upper plate of the UAV is connected to the top plate through four vertical support pillars, and the lower part of the upper plate of the UAV is connected to the flight module through a lower pad.

[0010] As the preferred technical solution of the present application, the flight module of the inspection drone includes four flight arms, each of which is equipped with a motor at the outer end. The motor is connected to the flight arm through a motor mounting upper plate and a motor mounting lower plate. The motor shaft of the motor is connected to the fan blades. A cross connector is installed on the motor mounting upper plate. The fan blade protector is fixed to the fan blade protection upper plate through the cross connector to protect the fan blades. At the same time, it plays a role in optimizing airflow management, reducing aerodynamic noise, and improving flight efficiency.

[0011] As the preferred technical solution of the present application, the landing module of the inspection UAV includes a pair of landing brackets, which are connected to the flight arm through a lower plate. The landing brackets are made of high-temperature resistant and impact-resistant materials. The brackets are provided with a lower landing bracket with strong grip, which can stably land on various complex terrains.

[0012] As a preferred technical solution of the present application, the lower plate of the inspection drone is connected to an identification module, and the identification module includes a dual-spectrum camera. The dual-spectrum camera is connected to the lower plate via a camera connector. The dual-spectrum camera is responsible for transmitting camera data to the embedded processor, and the embedded processor locates and describes the fire source through a built-in fire source identification algorithm.

[0013] As the preferred technical solution of the present application, the outer shell of the inspection drone is made of flame-retardant PC alloy material, and the inner surface close to the heat source is coated with a ceramic-based high-temperature resistant coating to protect the electronic components of the drone from damage caused by direct flame contact or high-temperature radiation generated by forest fires.

[0014] As the preferred technical solution of the present application, the inspection drone can maintain real-time communication with the control center through the WiFi module when performing fire source detection tasks, and can instantly transmit important information such as the location of the fire source and the fire status. At the same time, the control center can also manually control the flight of the drone.

[0015] A method for operating a patrol drone for forest fire inspection includes the following steps:

[0016] S1, the inspection drone cruises over the designated area, the dual-spectrum camera performs real-time monitoring, and transmits data to the embedded processor;

[0017] S2. The embedded processor analyzes the data and determines whether there is a fire source according to a preset fire source identification algorithm. If there is a fire source, the flight altitude and speed of the drone can be adjusted according to the environment and flight status to optimize the monitoring effect, and the fire source location, fire status and other information will be immediately transmitted to the control center in real time through the WiFi module.

[0018] As a preferred technical solution of the present application, during the cruising process of the inspection UAV, the embedded processor can adjust the flight altitude and speed of the UAV according to the environment and flight status to optimize the monitoring effect.

[0019] Beneficial Effects of the Invention

[0020] 1. In the present invention, the quadcopter drone uses a flame recognition algorithm based on deep learning technology to automatically identify the fire source, and enables the drone to have stable flight capabilities in complex environments, greatly improving the speed and accuracy of fire source detection, and breaking through the limitations of traditional drone working modes.

[0021] 2. Compared with the prior art, the UAV design of the present invention not only adapts to various environmental conditions and reduces the interference of manual operation, but also significantly improves the efficiency and safety of fire inspections, is beneficial to the early detection and timely control of forest fires, and greatly enhances the protection capacity of forest resources.

[0022] 3. The real-time communication function of the drone can effectively assist the ground control center to better allocate rescue resources, conduct scientific analysis and decision-making on the fire situation, and reduce resource waste and rescue delays caused by delayed and inaccurate information. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the overall structure of a patrol drone designed for forest fires provided in this application;

[0024] Figure 2 A schematic diagram of the structure of a control module for a patrol drone used for forest fires provided in this application;

[0025] Figure 3 A schematic diagram of the design structure of a drone flight module for forest fire inspection provided in this application;

[0026] Figure 4 A schematic diagram of the upward design structure of a patrol drone identification module for forest fires provided in this application; Figure 5 A schematic diagram of the structure of a wireless communication module of a patrol drone for forest fire inspection provided in this application.

[0027] Markings in the figure:

[0028] 1. Main control support board; 10. Embedded processor; 11. Upper pad; 12. UAV upper board; 13. Support column; 14. Top plate; 15. Lower pad; 2. Flight arm; 20. Motor; 21. Motor mounting upper plate; 22. Motor mounting lower plate; 23. Fan blade; 24. Cross connector; 25. Fan blade protector; 3. Landing bracket; 30. Lower plate; 31. Landing lower bracket; 4. Dual-spectrum camera; 40. Camera connector; 5. WiFi module. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] As described in the background art, existing drones in the prior art mainly rely on manual operation to identify fire sources, and may work abnormally or fail in complex and extreme environments.

[0031] In order to solve this technical problem, the present invention provides a design of a patrol drone for forest fires, which is applied in the field of forest fire prevention.

[0032] When the UAV performs accurate flame recognition, it mainly uses a flame recognition algorithm based on deep learning technology, especially an improved convolutional neural network (CNN). The implementation algorithm mainly includes the following steps:

[0033] S1. Data collection: The dual-spectrum camera equipped with the drone captures real-time images of the forest farm, including infrared images I IR Used to detect heat sources and visible light images I vis Used to obtain detailed visual features of the scene.

[0034] S2. Preprocessing: Standardize the image data to reduce the interference of lighting changes and background noise.

[0035]

[0036] Among them, μ and σ represent the mean and standard deviation of the image respectively.

[0037] S3. Feature extraction and fusion: The pre-trained CNN model f(·; θ) is used to extract key features from the processed images, and the two types of features are fused to improve the recognition accuracy of flames.

[0038] Feature extraction:

[0039]

[0040] Feature Fusion:

[0041] ·V fusion =V IR ||V vis

[0042] where || represents the feature fusion operation, θ IR and θ vis Represents the parameters of the CNN model trained for visible light and infrared light images.

[0043] S4, flame recognition: using the fused V fusion and flame classifier C(·;ω) to assess the likelihood of a fire source.

[0044] ·P fire =C(V fusion ;ω)

[0045] Where P fire is the fire source probability predicted by the model, and ω is the parameter of the classifier.

[0046] S5. Probability prediction: If P fire If it exceeds a predetermined value τ, it is recognized that a flame exists.

[0047] The parameters of the CNN model f(·; θ) and the flame classifier C(·; ω) are trained on a large number of labeled datasets through a supervised learning algorithm. The training process involves an optimization algorithm to minimize the error function J(θ, ω), usually taking the cross entropy loss.

[0048] Error function:

[0049]

[0050] Among them, N is the total number of training samples, y i is the true label of the i-th sample.

[0051] After identifying the fire source, the drone can automatically adjust its flight path based on an advanced path planning algorithm to stably hover 50 meters above the fire source for continuous monitoring. The algorithm is based on a PID (proportional-integral-differential) controller to optimize the drone's navigation. The main steps are as follows:

[0052] 1. Target setting: The algorithm first determines the location of the fire source as the target point:

[0053] P target =(x target ,y tartet, z target +50), where z target +50 ensures the drone is 50 meters above the fire source.

[0054] 2. Real-time positioning: The drone continuously updates its current position P through the GPS sensor in the embedded processor curent =(x current ,y current , z current ).

[0055] 3. Error calculation: Calculate the position error ΔP=P between the current position of the drone and the target position target -P current .

[0056] 4. PID control: Based on the position error ΔP, the PID controller adjusts the velocity vector V of the drone adjusted To minimize errors and achieve smooth flight.

[0057] Proportional term P: P = K p ΔP

[0058] Integral term I: I = I previous +Ki·ΔP·Δt

[0059] Differential term D: D = K d ·(ΔP-ΔP previous ) / Δt

[0060] Where K p , K i and K d are the proportional, integral and differential coefficients of the PID controller respectively, and Δt represents the time interval.

[0061] 5. Speed ​​adjustment: add the adjustment values ​​obtained by PID control to obtain the adjusted speed vector V of the drone adjusted =P+I+D.

[0062] 6. Implement navigation: The drone follows V adjusted Adjust its flight direction and speed until it reaches the position above the target for continuous monitoring, where the navigation termination condition is the position P of the UAV. current and the target position P target When the difference is lower than the preset threshold, the UAV is considered to have arrived at the location and starts to perform the monitoring mission.

[0063] For details, please refer to Figure 1, a patrol drone for forest fires, characterized in that it includes a main control module, a wireless communication module is arranged above the main control module, and the main control module is connected to the flight module below, the flight module is connected to the landing module below, and the landing module is connected to the identification module below, the main control module includes a main control support plate 1, an embedded processor 10 is placed above the main control support plate 1, the lower end of the main control support plate 1 is connected to the drone upper plate 12 through four upper pads 11, the drone upper plate is connected to the top plate 14 through four vertical support pillars 13, the lower part of the drone upper plate is connected to the flight module through a lower pad 15, the flight module includes four flight arms 2, each of the flight arms 2 is equipped with a motor 20 at the outer end, the motor 20 is connected to the flight arm through a motor mounting upper plate 21 and a motor mounting lower plate 22, the motor shaft of the motor 20 is connected to the fan blade 23, the motor A cross connector 24 is installed on the upper plate 21 of the drone, and a blade protector 25 is fixed to the blade protection upper plate 21 through the cross connector 24 to protect the blades 23. The landing module includes a pair of landing brackets 3, and the landing brackets 3 are connected to the flight arm 2 through a lower plate 30. The landing brackets 3 are made of high temperature resistant and impact resistant materials. The brackets are provided with landing lower brackets 31 with strong grip, which can stably land on various complex terrains. The lower plate 30 is connected to the identification module below, and the identification module includes a dual-spectrum camera 4, and the dual-spectrum camera 4 is connected to the lower plate 30 through a camera connector 40. The dual-spectrum camera 4 is responsible for transmitting camera data to the embedded processor 10. The embedded processor 10 locates and describes the fire source through a built-in fire source identification algorithm. There is also a WiFi module placed on the upper part of the drone.

[0064] like Figure 2 As shown in the figure, it is the control module of the inspection drone. As one of the core components of the drone, the control module is mainly composed of the main control support board and the embedded processor. The main control support board provides a stable assembly platform for the embedded processor, and the embedded processor connects and controls all the working modules of the drone. It is responsible for receiving and processing flight instructions from the remote control or ground control station, and controls the speed and steering of the motor according to the received instructions to adjust the flight attitude. In addition, it can process flight data in real time through the built-in system software, such as battery power, flight altitude, positioning and other information, to ensure the safety and stability of the drone's flight.

[0065] like Figure 3The figure shows the flight module of the inspection drone. The flight module is the key part of the drone responsible for taking off, flying and landing. It includes four flight arms, and each flight arm is equipped with a motor at the outer end. The motor is connected to the flight arm through the motor mounting upper plate and the motor mounting lower plate. The motor shaft of the motor is connected to the fan blade. The motor mounting upper plate is equipped with a cross connector. The fan blade protector is fixed to the fan blade protection upper plate through a cross connector to protect the fan blade. The fan blade is controlled by the motor to rotate to generate lift to help the drone take off. During the flight, the control system can achieve the drone's hovering, forward and backward, left and right movement, rise and fall, and left and right rotation and other flight actions by changing the speed of the four motors. All components are fixed on the flight arm to form a stable and easy-to-maintain frame. The design and working principle of the flight module enable the drone to achieve stable and flexible control in different flight environments.

[0066] like Figure 4 The figure shows the identification module of the inspection drone. In the process of forest inspection, the dual-spectrum camera carried by the drone becomes a key part. By capturing two types of images, visible light and thermal imaging, it can provide more comprehensive and accurate environmental information. It is this dual-spectrum feature that allows the drone to monitor the ground status in real time and transmit the data to the embedded processor. The built-in fire source identification algorithm of the embedded processor will immediately analyze the visible light and thermal imaging data. When the fire source is identified, the specific location of the fire source and the status of the fire are sent to the control center in real time through the wireless communication module. At the same time, this embedded processor adjusts the flight altitude and speed of the drone in real time according to the analysis results of the surrounding environment and the flight status of the drone to optimize the effect of fire source monitoring.

[0067] like Figure 5 As shown, this is the wireless communication module of the drone. When the drone performs cargo source detection tasks, it can maintain real-time communication with the control center through the WiFi module, even transmitting important information including the location of the fire source and the severity of the fire. At the same time, the control center can also manually control the flight of the drone.

[0068] Practical example: The design and use of a patrol drone for forest fires includes the following steps:

[0069] Step 1: After the above-mentioned quadcopter is assembled, the drone is first subjected to necessary pre-flight tests, including but not limited to self-inspection of each functional module, comprehensive inspection of the flight power system, preheating and calibration of the camera system, and connection of the communication module and establishment of the communication link.

[0070] Step 2: After the pre-flight test is passed, the drone enters the launch preparation stage. The preheated camera will start the preset scanning mode and begin to continuously acquire and send image information of the surrounding environment. The drone will autonomously generate startup and flight plans based on this information. After the flight plan is generated, the drone will start the flight program to take off and enter the inspection mode. During the flight, the drone will determine whether there is a fire source based on the real-time image data. If there is a fire source, it will automatically switch to the fire source tracking mode and send the fire source information back to the ground control center.

[0071] Step 3: In the fire source tracking mode, the drone will continue to move towards the fire source, using the dual-spectrum camera to obtain more detailed fire source data and send it to the ground control center in real time. While obtaining the fire source data, the drone will maintain a certain safe distance so that it will not be directly affected by the fire source.

[0072] Step 4: In long-term fire source tracking or inspection mode, if the battery power is lower than the preset value, the drone will automatically start the return program and return to the launch site. At the same time, ground staff can also adjust the drone's mission at any time through the remote control.

[0073] Obviously, the embodiments described above are only some embodiments of the present invention, not all embodiments, but do not limit the patent scope of the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for a person skilled in the art to modify the technical solutions described in the aforementioned specific embodiments, or to perform equivalent replacements on some of the technical features. Any equivalent structure made using the contents of the specification and drawings of the present invention, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present invention.

Claims

1. A patrol drone for forest fires, characterized in that It includes a main control module. A wireless communication module is arranged on the upper part of the main control module, which is connected to a flight module at the lower part. The flight module is connected to a landing module at the lower part, and the landing module is connected to an identification module at the lower part.

2. The inspection drone according to claim 1, characterized in that The main control module comprises a main control support plate (1), an embedded processor (10) is placed above the main control support plate (1), the lower end of the main control support plate (1) is connected to an upper plate (12) of a drone via four upper pads (11), the upper plate of the drone is connected to a top plate (14) via four vertical support pillars (13), and the lower part of the upper plate of the drone is connected to a flight module via a lower pad (15).

3. The inspection drone according to claim 2, characterized in that The flight module comprises four flight arms (2), each of the flight arms (2) being provided with a motor (20) at its outer end, the motor (20) being connected to the flight arm via a motor mounting upper plate (21) and a motor mounting lower plate (22), the motor shaft of the motor (20) being connected to a fan blade (23), the motor mounting upper plate (21) being provided with a cross connecting piece (24), a fan blade protector (25) being fixed to the fan blade protection upper plate (21) via the cross connecting piece (24), and being used for protecting the fan blade (23), and at the same time playing the role of optimizing airflow management, reducing aerodynamic noise, and improving flight efficiency.

4. The inspection drone according to claim 3, characterized in that The landing module comprises a pair of landing brackets (3), wherein the landing brackets (3) are connected to the flight arm (2) via a lower plate (30), the landing brackets (3) are made of high temperature resistant and impact resistant materials, and the landing brackets (3) are provided with a lower landing bracket (31) with strong grip, so that the landing can be stably landed on various complex terrains.

5. The inspection drone according to claim 1, characterized in that The bottom of the landing module is connected to the identification module, which includes a dual-spectrum camera (4). The dual-spectrum camera (4) is connected to the landing module via a camera connector (40). The dual-spectrum camera (4) is responsible for transmitting camera data to the embedded processor (10). The embedded processor (10) locates and describes the fire source through a built-in fire source identification algorithm; or performs flight path planning and automatically adjusts its flight path to stably hover above the fire source for continuous monitoring.

6. The inspection drone according to claim 5, characterized in that The fire source identification algorithm includes the following steps: S1. Data collection: The dual-spectrum camera equipped with the drone captures real-time images of the forest farm, including infrared images I IR Used to detect heat sources and visible light images I vis To obtain detailed visual features of the scene; S2. Preprocessing: Standardize the image data by the following formula to reduce the interference of illumination changes and background noise: in, represents the normalized infrared image, represents the normalized visible light image; μ and σ represent the mean and standard deviation of the image, respectively; IR and vis are the subscripts of infrared and visible light images, respectively; S3. Feature extraction and fusion: Use the pre-trained CNN model f(·; θ) to extract key features from the processed image, and fuse the two types of features to improve the recognition accuracy of flames: Feature extraction: Among them, V IR represents the extracted infrared image features, V vis represents the extracted visible light image features; θ IR and θ vis Represents the parameters of the CNN model trained for visible light and infrared light images; Feature Fusion: V fusion =V IR ||V vis Among them, V fusion represents the fused features, || represents the feature fusion operation; S4, flame recognition: using the fused V fusion and the flame classifier C(·;ω) to assess the likelihood of a fire source: P fire =C(V fusion ;ω) Where P fire is the probability of fire source predicted by the model, and ω is the parameter of the classifier; S5. Probability prediction: If P fire If it exceeds the predetermined value τ, it is recognized as the presence of flame; The parameters of the CNN model f(·; θ) and the flame classifier C(·; ω) are trained on a large number of labeled data sets through a supervised learning algorithm. The training process involves an optimization algorithm to minimize the error function J(θ, ω), which is usually taken as the cross entropy loss. Error function: Among them, N is the total number of training samples, y i is the true label of the i-th sample.

7. The inspection drone according to claim 5, characterized in that The flight path planning algorithm includes the following steps: S1. Target setting: The algorithm first determines the fire source location as the target point: P target =(x target ,y target ,z target +50) Among them, (x target ,y target , z target ) represents the fire source location coordinates, z target +50 ensures the drone is 50 meters above the fire source; S2, real-time positioning: The drone continuously updates its current position P through the GPS sensor in the embedded processor current =(x current ,y current , z current ); S3. Error calculation: Calculate the position error ΔP=P between the current position of the drone and the target position target -P current ; S4, PID control: Based on the position error ΔP, the PID controller adjusts the velocity vector V of the drone adjusted To minimize errors and achieve smooth flight: Proportional term P: P = K p ΔP Integral term I: I = I previous +K i ·ΔP·Δt Differential term D: D = K d ·(ΔP-ΔP previous ) / Δt Where K p , K i and K d are the proportional, integral and differential coefficients of the PID controller respectively, and Δt represents the time interval; S5, speed adjustment, add the adjustment value obtained by PID control to get the adjusted speed vector V of the drone adjusted =P+I+D; S6. Implement navigation: The drone follows V adjusted Adjust its flight direction and speed until it reaches the position above the target for continuous monitoring, where the navigation termination condition is the position P of the UAV. current and the target position P target When the difference is lower than the preset threshold, the UAV is considered to have arrived at the location and starts to perform the monitoring mission.

8. The inspection drone according to claim 5, characterized in that The outer shell of the inspection drone is made of flame-retardant PC alloy material, and the inner surface close to the heat source is coated with a ceramic-based high-temperature resistant coating to protect the electronic components of the drone from direct flame contact or high-temperature radiation caused by forest fires.

9. The inspection drone according to claim 5, characterized in that When the inspection drone performs a fire source detection mission, it can maintain real-time communication with the control center through the WiFi module (5), and instantly transmit important information including the location of the fire source and the fire status. At the same time, the control center can also manually control the flight of the drone.

10. An operation method of a patrol drone applied to forest fires, based on the patrol drone according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1, the inspection drone cruises over a designated area, the dual-spectrum camera (4) performs real-time monitoring, and transmits data to the embedded processor (10); S2. The embedded processor (10) analyzes the data and determines whether a fire source exists according to a preset fire source identification algorithm. If a fire source exists, the fire source location and fire status information are immediately transmitted to the control center in real time via the WiFi module (5).

Images