Water curtain and air curtain protection unmanned aerial vehicle system and control method thereof

Abstract

The application discloses a water curtain and air curtain protection unmanned aerial vehicle system and a control method thereof, and relates to the technical field of fire-fighting equipment. The system comprises a machine body, a ground unmanned aerial vehicle control system, an infrared thermal imager arranged on one side of the machine body, a water curtain forming device arranged below the machine body and an air curtain forming device. The water curtain forming device comprises a water curtain nozzle controller in communication connection with a water pump, the water pump is connected with a water tank pipeline and the water pump is connected with a water curtain nozzle pipeline. The air curtain forming device comprises an air curtain nozzle controller in communication connection with a gas pump, the gas pump is connected with a high-pressure gas generating device pipeline and the gas pump is connected with an air curtain nozzle pipeline. The application adopts a water curtain and air curtain coupling spraying technology. The water curtain reduces the ambient temperature by evaporative heat absorption and blocks high-temperature radiation. The air curtain prevents smoke diffusion. The double protection synergistic effect effectively controls the human body surface temperature within a bearable temperature range, and greatly prolongs the safety rescue window period.

Description

Technical Field

[0001] This invention relates to the field of fire-fighting equipment technology, and in particular to a water curtain and air curtain protection drone system and its control method. Background Technology

[0002] With the acceleration of urbanization, high-rise buildings have become the mainstream form of cities. The dense vertical shafts such as pipe shafts and cable shafts inside high-rise buildings are prone to creating a "chimney effect" in the event of a fire, causing smoke to spread vertically at an extremely fast speed. The dual threat of high-temperature smoke and toxic gases forces people to stay at windows or balconies, while traditional fire ladder trucks are limited by their lifting height and response time, making it difficult to provide rapid rescue.

[0003] Currently, in high-altitude rescue technologies, drones are mainly used for fire reconnaissance or material delivery; traditional aerial ladder trucks are often unable to reach the rescue site quickly due to road conditions; although descent devices can be used for personnel escape, they need to be installed in advance and are difficult to deal with sudden fires; helicopter rescue is affected by dense smoke at the fire scene and has extremely high flight risks, which limits its application in complex fire scene environments. Summary of the Invention

[0004] Therefore, it is necessary to provide a water curtain and air curtain protection drone system and its control method to address the above-mentioned technical problems.

[0005] This invention provides a water curtain and air curtain protection drone system, comprising:

[0006] Airframe, ground-based UAV control system, infrared thermal imager located on one side of the airframe, and water curtain forming device and air curtain forming device located below the airframe;

[0007] The water curtain forming device includes: a water tank, a water pump, a water curtain nozzle controller, and water curtain nozzles; the water curtain nozzle controller is communicatively connected to the water pump, the water pump is connected to the water tank via a pipe, and the water pump is connected to the water curtain nozzle via a pipe.

[0008] The air curtain forming device includes: a high-pressure gas generating device, an air pump, an air curtain nozzle controller, and air curtain nozzles; the air curtain nozzle controller is communicatively connected to the air pump, the air pump is connected to the high-pressure gas generating device via a pipeline, and the air pump is connected to the air curtain nozzle via a pipeline.

[0009] Optionally, the bottom of the machine body is fixedly connected to a mounting plate by multiple brackets, and the bottom of the mounting plate is equipped with a spraying device;

[0010] The injection device includes: a rear seat, a water-air distribution mechanism, a connecting frame, and an annular injection cylinder; the injection device is connected to the bottom of the mounting plate via the connecting frame.

[0011] The annular spray nozzle is connected to the water-air distribution mechanism, which has a built-in air curtain temporary storage-distribution unit and a water curtain temporary storage-distribution unit.

[0012] The annular spray cylinder is equipped with an annular spray assembly, which includes an air curtain nozzle, a water curtain nozzle, a gas pressure relief port, and a liquid pressure relief port; the air curtain nozzle is connected to the air curtain temporary storage-distribution unit, and the water curtain nozzle is connected to the water curtain temporary storage-distribution unit.

[0013] The gas pressure relief port and the liquid pressure relief port are arranged along the axial direction of the annular spray cylinder and are close to the middle area where the axis is located, and are respectively connected to the air curtain temporary storage-distribution unit and the water curtain temporary storage-distribution unit;

[0014] The water curtain nozzle controller is connected to the water curtain temporary storage-distribution unit via the water curtain nozzle control line, and the air curtain nozzle controller is connected to the air curtain temporary storage-distribution unit via the air curtain nozzle control line.

[0015] Optionally, the water pump is connected to one end of the No. 1 water pipe and one end of the No. 2 water pipe respectively. The other end of the No. 1 water pipe extends into the water tank, and the other end of the No. 2 water pipe is connected to the water curtain temporary storage-distribution unit.

[0016] Water pumps, No. 1 water pipe and No. 2 water pipe are used to transfer water from the water tank to the water curtain temporary storage-distribution unit;

[0017] The air pump is connected to one end of the No. 1 air pipe and one end of the No. 2 air pipe respectively. The other end of the No. 1 air pipe is connected to the high-pressure gas generating device, and the other end of the No. 2 air pipe is connected to the air curtain temporary storage-distribution unit.

[0018] The air pump, air pipe No. 1, and air pipe No. 2 are used to transfer gas from the high-pressure gas generating device to the air curtain temporary storage-distribution unit.

[0019] Optionally, a bracket is also provided below the spraying device, and the bracket is fixedly connected to the bottom of the mounting plate by multiple pillars;

[0020] Brackets are used to support the drone while it stands upright.

[0021] Optionally, the high-pressure gas generating device includes: multiple high-pressure gas cylinders, gas cylinder solenoid valves, and a high-pressure gas concentrator, with one end of the high-pressure gas concentrator connected to the high-pressure gas cylinders and the other end connected to a gas pump.

[0022] Gas cylinder solenoid valve is used to control the opening and closing of high-pressure gas cylinders;

[0023] Multiple high-pressure gas cylinders are used to generate gas;

[0024] High-pressure gas concentrators are used to centrally transmit gas generated from multiple high-pressure gas cylinders to a gas pump.

[0025] Optionally, the top of the aircraft is equipped with an information receiving and transmission terminal, which is used to receive image data from the infrared thermal imager and transmit it to the ground UAV control system, and to receive water curtain spray parameters and air curtain spray parameters fed back by the ground UAV control system and transmit them to the water curtain nozzle controller and air curtain nozzle controller.

[0026] Optionally, the information receiving and transmitting terminal includes a wireless transceiver module, a multi-protocol data processing unit, and an embedded control chip;

[0027] The wireless transceiver module is used to receive infrared image data and transmit it to the ground UAV control system, as well as to transmit the digital signals fed back by the ground UAV control system to the water curtain nozzle controller and the air curtain nozzle controller.

[0028] A multi-protocol data processing unit is used to classify fire data according to the importance of the infrared image data.

[0029] Embedded control chip for controlling water curtain and air curtain protection drone systems.

[0030] This invention also provides a control method for a water curtain and air curtain protective drone system, comprising:

[0031] Infrared image data of the fire scene is acquired using an infrared thermal imager;

[0032] The ground-based UAV control system performs image recognition on infrared image data to obtain flame boundary detection results, and predicts the temperature experienced by trapped personnel based on the flame boundary detection results, so as to determine the air curtain spray parameters and water curtain spray parameters.

[0033] The water flow rate of the water pump drawn from the water tank is controlled according to the water curtain spray parameters, and the water curtain nozzles are controlled by the water curtain nozzle controller to convert the water into a water curtain; the amount of pressurized gas generated by the high-pressure gas generating device and the air pump is controlled according to the air curtain spray parameters, and the pressurized gas is converted into an air curtain by the air curtain nozzle controller to form a protective barrier to protect people trapped at the fire scene.

[0034] Optionally, image recognition is performed on the infrared image data using a flame recognition model, specifically including:

[0035] The improved YOLOv5 convolutional neural network consists of an input layer, a convolutional layer, an activation layer, a pooling layer, a fully connected layer, and an output layer connected in sequence.

[0036] The image data of the fire scene is divided into multiple sub-regions, and the grayscale histogram of each sub-region is determined based on the following formula:

[0037] ;

[0038] in, nk grayscale k Quantity, N This represents the total number of pixels in the sub-block.

[0039] By progressively extracting low-level features from the grayscale histograms of each sub-region through convolutional layers, the edge texture features of the fire are obtained.

[0040] The shape features of the fire are obtained by feature mapping of the edge texture features of the fire through activation layers;

[0041] The shape features of the fire are pooled using a pooling layer to extract semantic information about the fire and obtain flame boundary detection results.

[0042] The flame boundary detection results are mapped to class probabilities through a fully connected layer to determine the loss function;

[0043] By updating the weights and biases of the convolutional kernels using a chain rule in the output layer to minimize the loss function, an improved YOLOv5 convolutional neural network is trained to obtain a flame recognition model.

[0044] Optionally, the temperature experienced by trapped personnel can be predicted based on the flame boundary detection results to determine the air curtain and water curtain spray parameters, specifically including:

[0045] The maximum temperature experienced by the human body is determined based on the flame boundary detection results using the following formula:

[0046] ;

[0047] ;

[0048] in, T 1 represents the highest temperature the human body can experience. T 0 is the normal human body temperature. T max The highest temperature of the flame in the flame boundary detection results. F 12 The effective radiation ratio;

[0049] When the highest temperature experienced by the human body exceeds a set threshold, the water curtain spray parameters are determined based on the following formula:

[0050] ;

[0051] ;

[0052] in, v 0 Let the initial velocity of the water flow be _____. It is the acceleration due to gravity. L The length of the water curtain coverage. The angle of the water curtain spray. P For water pump pressure, For the density of water, h For Yang Cheng;

[0053] The air curtain injection parameters are determined based on the following formula:

[0054] ;

[0055] in, The air curtain outlet velocity, The density of water, The density is the gas density.

[0056] The water curtain and air curtain protection drone system and its control method provided in this invention have the following advantages compared with the prior art:

[0057] This invention employs a water curtain and air curtain coupled spraying technology. The water curtain and air curtain are delivered to independent units of the annular spray cylinder through a water-air distribution mechanism. They are then sprayed synchronously by the air curtain nozzles and water curtain nozzles, achieving a combined protection of water mist cooling and airflow smoke isolation. Furthermore, the water curtain reduces the ambient temperature by evaporating and absorbing heat, while blocking high-temperature radiation, and the air curtain isolates the spread of smoke. Together, they form a dual protective layer of "cooling + smoke isolation," effectively controlling the human body surface temperature within a tolerable range and significantly extending the safe rescue window.

[0058] Furthermore, the water curtain forming device and air curtain forming device in this invention adopt a modular layout, which facilitates quick replacement and maintenance, ensuring the reliability and continuity of the equipment in emergency missions. It also overcomes the limitations of ladder trucks and helicopters due to dense smoke by utilizing the high mobility of UAVs, which can quickly reach any height of high-rise buildings, thus providing flexible and reliable technical support for high-rise fire rescue. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of a water curtain and air curtain protection drone system provided in one embodiment;

[0060] Figure 2 This is a schematic diagram of the spray device of a water curtain and air curtain protection drone system provided in one embodiment;

[0061] Figure 3 This is a schematic diagram of the structure of an annular jet tube of a water curtain and air curtain protection drone system provided in one embodiment;

[0062] Figure 4 This is a flowchart illustrating a control method for a water curtain and air curtain protection drone system provided in one embodiment.

[0063] The components include: 1. Airframe; 2. Wings; 3. Water curtain nozzle controller; 4. Air curtain nozzle controller; 5. Infrared thermal imager; 6. Information receiving and transmission terminal; 7. Bracket; 8. Mounting plate; 9. Water tank; 10. High-pressure gas generating device; 10-1. High-pressure gas cylinder; 10-2. Gas cylinder solenoid valve; 10-3. High-pressure gas central pipe; 11. Water pump; 12. Air pump; 13. Spraying device; 13-1. Rear seat; 13-2. Water-air distribution mechanism; 13-2-1. Air curtain temporary storage-distribution unit, 13-2-2, water curtain temporary storage-distribution unit; 13-3, connecting frame; 13-4, annular spray tube, 13-4-1, air curtain nozzle, 13-4-2, water curtain nozzle, 13-4-3, gas pressure relief port, 13-4-4, liquid pressure relief port; 14, No. 1 water pipe, 15, No. 2 water pipe, 16, No. 1 air pipe, 17, No. 2 air pipe, 18, water curtain nozzle control line, 19, air curtain nozzle control line, 20, bracket, 21, support column. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0065] This invention provides a water curtain and air curtain protection drone system, the system comprising:

[0066] Airframe 1, ground unmanned aerial vehicle control system, infrared thermal imager 5 located on one side of the exterior of airframe 1, and water curtain forming device and air curtain forming device located below airframe 1.

[0067] The water curtain forming device includes: a water tank 9, a water pump 11, a water curtain nozzle controller 3, and water curtain nozzles 13-4-2. The water curtain nozzle controller 3 is communicatively connected to the water pump 11, and the water pump 11 is pipe-connected to the water tank 9 and also pipe-connected to the water curtain nozzles 13-4-2.

[0068] The air curtain forming device includes: a high-pressure gas generating device 10, an air pump 12, an air curtain nozzle controller 4, and an air curtain nozzle 13-4-1. The air curtain nozzle controller 4 is communicatively connected to the air pump 12, and the air pump 12 is pipe-connected to the high-pressure gas generating device 10 and the air curtain nozzle 13-4-1.

[0069] Among them, the infrared thermal imager 5 is used to acquire infrared image data of the fire scene;

[0070] The ground-based unmanned aerial vehicle (UAV) control system is used to perform image recognition on infrared image data, obtain flame boundary detection results, and predict the temperature suffered by trapped personnel based on the flame boundary detection results, so as to determine the air curtain spray parameters and water curtain spray parameters.

[0071] The water curtain forming device is used to control the flow rate of water pump 11 from water tank 9 according to water curtain spray parameters, and to control water curtain nozzle 13-4-2 to convert water into a water curtain through water curtain nozzle controller 3.

[0072] The air curtain forming device is used to control the amount of pressurized gas generated by the high-pressure gas generating device 10 and the air pump 12 according to the air curtain injection parameters, and to control the air curtain nozzle 13-4-1 to convert the pressurized gas into an air curtain through the air curtain nozzle controller 4.

[0073] The specific implementation is as follows:

[0074] like Figure 1 As shown, a water curtain and air curtain protective drone system includes a body 1. Multiple wings 2 are evenly distributed around the body 1. A water curtain nozzle controller 3 and an air curtain nozzle controller 4 are mounted on the body 1. An infrared thermal imager 5 is mounted on one side of the body 1. An information receiving and transmission terminal 6 is mounted on the upper part of the body 1. A mounting plate 8 is fixedly connected to the bottom of the body 1 via multiple brackets 7. The mounting plate 8 is equipped with a water tank 9, a high-pressure gas generating device 10, a water pump 11, and an air pump 12. A spraying device 13 is located at the bottom of the mounting plate 8.

[0075] like Figure 2 and Figure 3 As shown, the spraying device 13 includes a rear seat 13-1, a water-air distribution mechanism 13-2, a connecting frame 13-3, and an annular spraying cylinder 13-4. The connecting frame 13-3 is used to connect the spraying device 13 to the bottom of the mounting plate 8. The annular spraying cylinder 13-4 communicates with the water-air distribution mechanism 13-2, which houses an air curtain temporary storage-distribution unit 13-2-1 and a water curtain temporary storage-distribution unit 13-2-2. The annular spraying cylinder 13-4 contains an annular spraying assembly, including an air curtain nozzle 13-4-1, a water curtain nozzle 13-4-2, a gas pressure relief port 13-4-3, and a liquid pressure relief port 13-4-4. The air curtain nozzle 13-4-1 communicates with the air curtain temporary storage-distribution unit 13-2-1, and the water curtain nozzle 13-4-2 communicates with the water curtain temporary storage-distribution unit 13-2-2. Gas pressure relief port 13-4-3 and liquid pressure relief port 13-4-4 are arranged along the axial direction of the annular spray cylinder and are located near the middle area of ​​the axis, respectively connected to the air curtain temporary storage-distribution unit 13-2-1 and the water curtain temporary storage-distribution unit 13-2-2. Water curtain nozzle controller 3 is connected to water curtain temporary storage-distribution unit 13-2-2 via water curtain nozzle control line 18, and is used to control the opening and closing of water curtain nozzle 13-4-2. Air curtain nozzle controller 4 is connected to air curtain temporary storage-distribution unit 13-2-1 via air curtain nozzle control line 19, and is used to control the opening and closing of air curtain nozzle 13-4-1.

[0076] Furthermore, water pump 11 is connected to one end of water pipe 14 and one end of water pipe 15, respectively. The other end of water pipe 14 extends into water tank 9, and the other end of water pipe 15 is connected to water curtain temporary storage-distribution unit 13-2-2. Water pump 11, water pipe 14, and water pipe 15 are used to transfer water from water tank 9 to water curtain temporary storage-distribution unit 13-2-2.

[0077] Air pump 12 is connected to one end of air pipe 16 and one end of air pipe 17. The other end of air pipe 16 is connected to high-pressure gas generating device 10, and the other end of air pipe 17 is connected to air curtain temporary storage-distribution unit 13-2-1. Air pump 12, air pipe 16, and air pipe 17 are used to transfer gas from high-pressure gas generating device 10 to air curtain temporary storage-distribution unit 13-2-1.

[0078] Furthermore, the information receiving and transmission terminal 6 is used to receive image data sent by the infrared thermal imager 5 in real time, and transmit the received image data signal to the ground UAV control system. The ground UAV control system processes the image data through the image processing program and outputs the water curtain-air curtain spray parameters. The information receiving and transmission terminal 6 receives the parameter data and transmits it to the water curtain nozzle controller 3 and the air curtain nozzle controller 4, and controls the pressure of the water pump and air pump and the water curtain and air curtain nozzles through the spray parameters.

[0079] Furthermore, the high-pressure gas generating device 10 includes multiple high-pressure gas cylinders 10-1, gas cylinder solenoid valves 10-2, and high-pressure gas concentrators 10-3; the gas cylinder solenoid valves 10-2 are used to control the opening and closing of the high-pressure gas cylinders 10-1, and one end of the high-pressure gas concentrators 10-3 is connected to the high-pressure gas cylinders 10-1, and the other end is connected to the gas pump 12.

[0080] Furthermore, the information receiving and transmission terminal 6 includes a wireless transceiver module, a multi-protocol data processing unit, and an embedded control chip. It supports 4G / 5G, WiFi, and anti-interference spread spectrum communication links. Its built-in data compression algorithm divides the infrared video frame into 64×64 coding tree units (CTUs), recursively subdividing them into smaller CU coding units. It utilizes the spatial correlation of adjacent blocks to reduce redundancy, applies discrete cosine transform to the residual data, and compresses high-frequency information through a quantization matrix to achieve H.265 encoding of the original infrared video. A dynamic priority scheduling protocol marks fire data as "urgent" (flame boundary, high-temperature point) and "normal" (background temperature field), allocating them to high / low priority queues respectively. Tag classification ensures that high-priority data packets occupy bandwidth resources first, guaranteeing low-latency transmission of fire data.

[0081] Wireless transceiver module: used to receive infrared image data and transmit it to the ground UAV control system, as well as the digital signals fed back by the ground UAV control system and transmit them to the water curtain nozzle controller 3 and the air curtain nozzle controller 4.

[0082] Multi-protocol data processing unit: Implements data protocol adaptation, encapsulation, and priority scheduling to ensure network compatibility and transmission efficiency. It is used to classify fire data according to the importance of the infrared image data content; for example, flame boundary coordinates are marked as high priority, and background temperature fields are marked as low priority. Appropriate scheduling ensures that high-priority data packets occupy bandwidth resources first.

[0083] Embedded control chip: Used to control the water curtain and air curtain protection drone system. It acts as the system's brain, coordinating hardware resources, executing core algorithms, and controlling the entire process. It leads the encoding process, including infrared image segmentation and compression. It manages the scheduling and coordination of multiple tasks.

[0084] Furthermore, a bracket 20 for supporting the drone to stand is provided below the jet device 13. The bracket 20 is fixedly connected to the bottom of the mounting plate 8 by multiple pillars 21.

[0085] Based on the same inventive concept, embodiments of the present invention also provide a control method for a water curtain and air curtain protective drone system, comprising the following steps:

[0086] Infrared image data of the fire scene was acquired using infrared thermal imager 5.

[0087] The ground-based UAV control system performs image recognition on infrared image data to obtain flame boundary detection results. Based on the flame boundary detection results, the temperature experienced by the trapped personnel is predicted to determine the air curtain and water curtain spray parameters.

[0088] The water flow rate of water pump 11 drawn from water tank 9 is controlled according to the water curtain spray parameters, and the water curtain nozzle 13-4-2 is controlled by the water curtain nozzle controller 3 to convert the water into a water curtain. The amount of pressurized gas generated by high-pressure gas generating device 10 and air pump 12 is controlled according to the air curtain spray parameters, and the pressurized gas is converted into an air curtain by air curtain nozzle 13-4-1 through air curtain nozzle controller 4 to form a protective barrier to protect people trapped at the fire scene.

[0089] like Figure 4 As shown, the specific implementation process is as follows:

[0090] S1. The infrared thermal imager 5 collects infrared image data of the fire scene and transmits the collected infrared image data to the ground drone control system in real time through the information receiving and transmission terminal 6.

[0091] S2. After receiving infrared image data, the ground-based UAV control system processes the infrared image data to obtain an enhanced infrared image. A YOLOv5 convolutional neural network is used to detect flame boundaries in the enhanced infrared image, and the flame boundary detection results are synchronously fed back to the UAV system for real-time optimization of spray parameters.

[0092] S3. The drone system sprays water and air curtains according to optimized spray parameters to form a protective barrier for trapped personnel.

[0093] Furthermore, the specific process of S2 is as follows:

[0094] The improved YOLOv5 convolutional neural network consists of an input layer, a convolutional layer, an activation layer, a pooling layer, a fully connected layer, and an output layer connected in sequence.

[0095] The image data of the fire scene is divided into multiple sub-regions, and the gray-level histogram of each sub-region is determined. Low-level features of the gray-level histograms of each sub-region are extracted progressively through convolutional layers to obtain the edge texture features of the fire. The edge texture features are then mapped using activation layers to obtain the shape features of the fire. Pooling layers are used to pool the shape features of the fire to extract semantic information, resulting in flame boundary detection. Fully connected layers map the flame boundary detection results to class probabilities to determine the loss function. Finally, the weights and biases of the convolutional kernels are updated using a chain rule in the output layer to minimize the loss function, training an improved YOLOv5 convolutional neural network to obtain a flame recognition model.

[0096] The implementation process is as follows:

[0097] S2.1 Divide the 640×640 image into 8×8 sub-regions, each sub-block being 80×80 pixels. Calculate the gray-level histogram for each sub-region. The formula for calculating the gray-level histogram is as follows:

[0098] ;

[0099] in, n k grayscale k Quantity, N This represents the total number of pixels in the sub-block.

[0100] The cumulative distribution function is: ;

[0101] The contrast gain is limited to a maximum of 2.0, and the calculation equation is as follows:

[0102] ;

[0103] Where clipLimit=2, if a certain gray level k number of pixels n ( k )> T If the excess portion is cropped, the extra pixels will be evenly distributed across all gray levels.

[0104] S2.2 The improved YOLOv5 convolutional neural network structure includes an input layer, convolutional layers, activation functions, pooling layers, fully connected layers, and an output layer. The input image passes through multiple convolution-activation-pooling layers, progressively extracting low-level features (edges, texture), mid-level features (shape), and high-level features (semantic information). The fully connected layer maps high-level features to class probabilities, calculates the loss function, backpropagates, and updates the convolutional kernel weights and biases using a chain rule, progressively training to minimize the loss function and obtain the flame recognition model.

[0105] Among them, the improvements to the YOLOv5 convolutional neural network structure include: 1. Adding a post-attention mechanism to the convolutional layers in the backbone network, generating channel weights through global average pooling, enhancing the feature response of flame-related channels, and locating the flame edge position through the fusion of max pooling and average pooling.

[0106] 2. Replace the activation function with SiLU The formula is as follows:

[0107] ;

[0108] for Sigmoid This function enhances the nonlinear expressive power and is more suitable for the complex distribution of flame characteristics.

[0109] The process of moving from low-level features to high-level features, from edges and textures to shapes, involves the following steps: extracting local edges and textures (such as the flickering edges of flames and the graininess of high-temperature areas); expanding the receptive field of neurons through multi-layer convolution (e.g., from 3×3 to 5×5) to capture more complex local patterns; reducing the resolution of the feature map through pooling operations (e.g., downsampling to 1 / 2); preserving salient features (such as the outline of flames); suppressing noise; focusing on the boundary region of the flame using an attention mechanism; and suppressing background interference such as smoke.

[0110] From intermediate to advanced features, from shape to semantic information, the steps are as follows: capture the overall distribution of the fire scene through a deep network, weight and fuse feature maps of different scales, combine local details with global semantics, and perform spatial weight allocation on multi-scale feature maps to dynamically enhance the semantic response of the core flame region. Finally, map the advanced features to the fire source temperature and bounding box coordinates.

[0111] S2.2-1. The training process uses the Adam optimizer with an initial learning rate of 3×10⁻⁶. −4 Cosine annealed to 10 −5 By adjusting the model parameters, i.e., the weights W and bias b Minimizing the loss function L is essentially a solution, expressed as follows:

[0112] ;

[0113] For each batch of data x Standardization is performed, including calculating batch statistics and normalization, with the following formulas:

[0114] ;

[0115] ;

[0116] in, This is the batch average. m This refers to the number of samples in the batch. This represents the batch variance. x i For the first in the input data i One sample; The data is the standardized sample data; It is a constant used for numerical stability.

[0117] S2.2-2, Scaling and translation to obtain the corresponding y :

[0118] ;

[0119] in, and Learnable;

[0120] S2.2-3. Add a weighted squared penalty term to the loss function. λ To enable optimizer settings such as weight_decay= λ To prevent overfitting, the Dropout technique is introduced, where each neuron is assigned a probability... p Randomly set to zero (e.g.) p =0.3, meaning a 30% probability of discarding; the output values ​​of neurons that are not discarded are calculated as 1 / 1- p Magnify; use training data to train step by step until the loss function is minimized to obtain an image recognition program based on a convolutional neural network;

[0121] S2.3. Based on the infrared image with flame boundaries output by the image recognition program based on convolutional networks, the fire source is defined as the radiating surface S1, and the area of ​​the trapped personnel is defined as the receiving surface S2. The human body temperature prediction model randomly emits multiple virtual "heat rays" towards the fire source area, tracks whether each ray penetrates the water curtain to reach the area of ​​the trapped personnel, calculates the heat blocking effect of different water mist densities (i.e., the effective radiation ratio), calculates the total temperature reaching the area of ​​the trapped personnel, and the system automatically optimizes the water spraying strategy.

[0122] The method to optimize the water spraying strategy is as follows:

[0123] When the total temperature in the area where the trapped personnel are located is detected to exceed the set threshold, the water mist density is increased. If the temperature exceeds the limit, the system gradually increases the blocking effect in fixed steps until the total temperature is below the threshold. After each adjustment, the system re-emits virtual "heat rays", calculates the updated effective radiation ratio and total temperature, and verifies the adjustment effect.

[0124] The formula for calculating the effective radiation ratio is as follows:

[0125] ;

[0126] The formula for determining the maximum temperature a human body can experience based on flame boundary detection results is as follows:

[0127] ;

[0128] in, T 1 represents the highest temperature the human body can experience. T 0 represents the normal human body temperature (309K). T max This represents the highest temperature of the flame in the infrared image. F 12 The effective radiation ratio.

[0129] Furthermore, the specific process of S3 is as follows:

[0130] S3.1. Obtain the appropriate water curtain density based on the calculation method in the human body temperature prediction model formula:

[0131] By monitoring the fire source temperature and current water mist density in real time, the effective radiation ratio is calculated using virtual ray simulation; if T If 1 is greater than the set threshold, the target water curtain density is calculated step by step according to S2.3.

[0132] S3.2. Calculate the water curtain spray parameters based on the distance from the drone nozzle to the area where the trapped personnel are located, and the following formula:

[0133] ;

[0134] in, v 0 represents the initial velocity of the water flow. It is the acceleration due to gravity. L The length of the water curtain coverage. The spray angle;

[0135] ;

[0136] in, P For the pressure of water pump 11, ρ w For the density of water, hThe pump head is calculated using this formula to determine the pressure required for the initial velocity of the water flow, thus controlling the pump.

[0137] S3.3 The jet velocity of the air curtain is calculated according to the following formula:

[0138] ;

[0139] in, The air curtain outlet velocity, The length of the water curtain coverage is determined by the distance from the fire. d With the angle of injection Joint decision, The density of water, The density is the gas density.

[0140] Example: The operator uses a remote control to control the drone to ascend to the vicinity of the floor where the trapped person is located, and at the same time turns on the infrared thermal imager 5, ensures that the water tank 9 is full of water and the gas cylinder is full of gas, and checks whether all water pipes and gas pipes are connected correctly.

[0141] When the drone flies to the vicinity of the floor where the trapped people are located, the ground drone control system begins to receive images from the infrared thermal imager 5. In the system, the flame range is marked by an image recognition program based on a convolutional neural network. Based on the processed infrared image with boundaries, the temperature of the people under different water curtain spray strategies is calculated according to the human body temperature prediction model to ensure that it reaches the built-in target of 60°C. The system also calculates the air curtain spray strategy that meets the air curtain envelopment requirements and transmits the spray strategy to the water curtain and air curtain spray nozzle controllers. The water curtain nozzle controller 3 opens the nozzles specified by the strategy and ensures the required flow and pressure output according to the received information. The air curtain nozzle controller 4 opens the nozzles specified by the strategy and ensures the required flow and pressure output according to the received air curtain spray strategy, creating a survival environment for the trapped people to wait for rescue personnel to rescue them from higher floors.

[0142] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A water curtain and air curtain protection system for unmanned aerial vehicles, characterized in that, include: The aircraft body (1), the ground unmanned aerial vehicle control system, the infrared thermal imager (5) located on one side of the exterior of the aircraft body (1), and the water curtain forming device and the air curtain forming device located below the aircraft body (1); The water curtain forming device includes: a water tank (9), a water pump (11), a water curtain nozzle controller (3), and a water curtain nozzle (13-4-2); the water curtain nozzle controller (3) is communicatively connected to the water pump (11), the water pump (11) is pipe-connected to the water tank (9), and the water pump (11) is pipe-connected to the water curtain nozzle (13-4-2); The air curtain forming device includes: a high-pressure gas generating device (10), an air pump (12), an air curtain nozzle controller (4), and an air curtain nozzle (13-4-1); the air curtain nozzle controller (4) is communicatively connected to the air pump (12), the air pump (12) is pipe-connected to the high-pressure gas generating device (10), and the air pump (12) is pipe-connected to the air curtain nozzle (13-4-1); The bottom of the body (1) is fixedly connected to the mounting plate (8) by multiple brackets (7), and the bottom of the mounting plate (8) is provided with a spraying device (13). The spraying device (13) includes: a rear seat (13-1), a water-air distribution mechanism (13-2), a connecting frame (13-3), and an annular spray cylinder (13-4); the spraying device (13) is connected to the bottom of the mounting plate (8) via the connecting frame (13-3). The annular spray tube (13-4) is connected to the water-air distribution mechanism (13-2), and the water-air distribution mechanism (13-2) has a built-in air curtain temporary storage-distribution unit (13-2-1) and a water curtain temporary storage-distribution unit (13-2-2). The annular spray cylinder (13-4) is equipped with an annular spray assembly, which includes an air curtain nozzle (13-4-1), a water curtain nozzle (13-4-2), a gas pressure relief port (13-4-3), and a liquid pressure relief port (13-4-4). The air curtain nozzle (13-4-1) is connected to the air curtain temporary storage-distribution unit (13-2-1), and the water curtain nozzle (13-4-2) is connected to the water curtain temporary storage-distribution unit (13-2-2). The gas pressure relief port (13-4-3) and the liquid pressure relief port (13-4-4) are arranged along the axial direction of the annular spray tube and close to the middle area where the axis is located, respectively connected to the air curtain temporary storage-distribution unit (13-2-1) and the water curtain temporary storage-distribution unit (13-2-2); The water curtain nozzle controller (3) is connected to the water curtain temporary storage-distribution unit (13-2-2) via the water curtain nozzle control line (18), and the air curtain nozzle controller (4) is connected to the air curtain temporary storage-distribution unit (13-2-1) via the air curtain nozzle control line (19).

2. The water curtain and air curtain protection drone system as described in claim 1, characterized in that, The water pump (11) is connected to one end of the No. 1 water pipe (14) and one end of the No. 2 water pipe (15), respectively. The other end of the No. 1 water pipe (14) extends into the water tank (9), and the other end of the No. 2 water pipe (15) is connected to the water curtain temporary storage-distribution unit (13-2-2). The water pump (11), the first water pipe (14) and the second water pipe (15) are used to transfer water from the water tank (9) to the water curtain temporary storage-distribution unit (13-2-2). The air pump (12) is connected to one end of the first air pipe (16) and one end of the second air pipe (17), respectively. The other end of the first air pipe (16) is connected to the high-pressure gas generating device (10), and the other end of the second air pipe (17) is connected to the air curtain temporary storage-distribution unit (13-2-1). The air pump (12), the first air pipe (16) and the second air pipe (17) are used to transfer gas from the high-pressure gas generating device (10) to the air curtain temporary storage-distribution unit (13-2-1).

3. The water curtain and air curtain protection drone system as described in claim 1, characterized in that, Below the spraying device (13) is a bracket (20), which is fixedly connected to the bottom of the mounting plate (8) by multiple support pillars (21); The bracket (20) is used to support the drone standing upright.

4. The water curtain and air curtain protection drone system as described in claim 1, characterized in that, The high-pressure gas generating device (10) includes: multiple high-pressure gas cylinders (10-1), gas cylinder solenoid valves (10-2) and high-pressure gas concentrator (10-3), one end of the high-pressure gas concentrator (10-3) is connected to the high-pressure gas cylinders (10-1) and the other end is connected to the gas pump (12). The gas cylinder solenoid valve (10-2) is used to control the opening and closing of the high-pressure gas cylinder (10-1); Multiple high-pressure gas cylinders (10-1) are used to generate gas; The high-pressure gas concentrator (10-3) is used to centrally transmit the gas generated by the multiple high-pressure gas cylinders (10-1) to the gas pump (12).

5. The water curtain and air curtain protection drone system as described in claim 1, characterized in that, The top of the body (1) is provided with an information receiving and transmission terminal (6), which is used to receive infrared image data from the infrared thermal imager (5) and transmit it to the ground UAV control system, and to receive water curtain spray parameters and air curtain spray parameters fed back by the ground UAV control system and transmit them to the water curtain nozzle controller (3) and the air curtain nozzle controller (4).

6. The water curtain and air curtain protection drone system as described in claim 5, characterized in that, The information receiving and transmission terminal (6) includes a wireless transceiver module, a multi-protocol data processing unit, and an embedded control chip; The wireless transceiver module is used to receive infrared image data and transmit it to the ground UAV control system and the digital signal fed back by the ground UAV control system and transmit it to the water curtain nozzle controller (3) and the air curtain nozzle controller (4). The multi-protocol data processing unit is used to classify fire data according to the importance of the infrared image data content; The embedded control chip is used to control the water curtain and air curtain protection drone system.

7. A control method for a water curtain and air curtain protective unmanned aerial vehicle system based on any one of claims 1-6, characterized in that, include: Infrared image data of the fire scene was acquired using an infrared thermal imager (5); The ground-based UAV control system performs image recognition on infrared image data to obtain flame boundary detection results, and predicts the temperature experienced by trapped personnel based on the flame boundary detection results, so as to determine the air curtain spray parameters and water curtain spray parameters. The water flow rate of the water pump (11) drawn from the water tank (9) is controlled according to the water curtain spray parameters, and the water curtain nozzle (13-4-2) is controlled by the water curtain nozzle controller (3) to convert the water into a water curtain; the amount of pressurized gas generated by the high pressure gas generating device (10) and the air pump (12) is controlled according to the air curtain spray parameters, and the pressurized gas is converted into an air curtain by the air curtain nozzle controller (4) to form a protective barrier to protect people trapped at the fire scene.

8. The control method for a water curtain and air curtain protective drone system as described in claim 7, characterized in that, Image recognition is performed on infrared image data using a flame recognition model, specifically including: The improved YOLOv5 convolutional neural network consists of an input layer, a convolutional layer, an activation layer, a pooling layer, a fully connected layer, and an output layer connected in sequence. The image data of the fire scene is divided into multiple sub-regions, and the grayscale histogram of each sub-region is determined based on the following formula: ； in, n k grayscale k Quantity, N This represents the total number of pixels in the sub-block. By progressively extracting low-level features from the grayscale histograms of each sub-region through convolutional layers, the edge texture features of the fire are obtained. The shape features of the fire are obtained by feature mapping of the edge texture features of the fire through activation layers; The shape features of the fire are pooled using a pooling layer to extract semantic information about the fire and obtain flame boundary detection results. The flame boundary detection results are mapped to class probabilities through a fully connected layer to determine the loss function; By updating the weights and biases of the convolutional kernels using a chain rule in the output layer to minimize the loss function, an improved YOLOv5 convolutional neural network is trained to obtain a flame recognition model.

9. The control method for a water curtain and air curtain protective drone system as described in claim 7, characterized in that, The step of predicting the temperature experienced by trapped personnel based on flame boundary detection results to determine air curtain and water curtain spray parameters specifically includes: The maximum temperature experienced by the human body is determined based on the flame boundary detection results using the following formula: ； ； in, T 1 represents the highest temperature the human body can experience. T 0 is the normal human body temperature. T max The highest temperature of the flame in the flame boundary detection results. F 12 The effective radiation ratio; When the highest temperature experienced by the human body exceeds a set threshold, the water curtain spray parameters are determined based on the following formula: ； ； in, v 0 Let the initial velocity of the water flow be _____. It is the acceleration due to gravity. L The length of the water curtain coverage. The angle of the water curtain spray. P For water pump pressure, For the density of water, h For Yang Cheng; The air curtain injection parameters are determined based on the following formula: ； in, The air curtain outlet velocity, The density of water, The density is the gas density.

Images