Three-state operation method and system for power transmission and distribution unmanned aerial vehicle spraying

Abstract

The application discloses a kind of power transmission and distribution unmanned plane spraying three-state operation method and system, the method includes the following steps: carrying out spraying experiment to spraying equipment, obtain visible light image to construct dataset, construct improved Faster R-CNN model and convolutional neural network model, data set and visible light image are respectively input in the model constructed and optimized, respectively obtain spraying equipment kind identification model and spraying quality evaluation model;Unmanned plane carries spraying mechanical arm and flies to the region to be sprayed after spraying, using spraying equipment kind identification model to identify the kind of equipment to be sprayed, obtain identification spraying equipment and carry out spraying, obtain spraying quality score, spraying quality score is detected by spraying quality evaluation model, when spraying quality score does not satisfy set threshold value, continue to spray, until spraying quality score reaches threshold value.The application can improve spraying speed and reduce manual operation by self-adaptive adjustment of flight control of spraying.

Description

Technical Field

[0001] This invention relates to the field of spraying power transmission lines and distribution lines, and particularly to a three-state operation method and system for spraying power transmission and distribution drones. Background Technology

[0002] In order to ensure the insulation performance of hardware, insulators and other areas of transmission lines and distribution networks, the existing technology uses manual modification of hardware, insulators and other areas of transmission lines and distribution networks. However, this modification method is costly and inefficient.

[0003] Existing technologies also employ drones to coat hardware and insulators of transmission and distribution lines with insulating coatings (such as RTV coatings) to improve insulation performance and extend service life. Heavy-duty drones can flexibly coat lines at high altitudes. However, current drone coating systems and devices only have a single mode, meaning they can only coat surfaces such as insulators. They lack surface cleaning and evaluation of coating effects, and current coating devices are inefficient, unable to be simultaneously applied to transmission and distribution lines, and lack safety in complex environments when coating power lines. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a three-state operation method and system for spraying power transmission and distribution drones, which solves the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a three-state operation method for spraying power transmission and distribution drones, comprising the following steps:

[0006] Acquire visible light images captured by a vision camera after a spraying experiment on a spraying equipment, and construct a dataset based on the visible light images;

[0007] The improved Faster R-CNN model was pre-trained using the dataset to obtain a model for identifying the types of spraying equipment.

[0008] Visible light images are imported into a convolutional neural network model for pre-training to obtain a coating quality evaluation model.

[0009] The process involves acquiring an image of the equipment to be sprayed, identifying the type of equipment using a spraying equipment type recognition model, determining the target modal operation from multiple modal operations based on the type of equipment, performing a spraying operation on the equipment according to the target modal operation, acquiring a visible light image of the equipment to be sprayed after the spraying operation, obtaining a spraying quality score by identifying the visible light image using a spraying quality evaluation model, and evaluating the spraying operation using the spraying quality score.

[0010] Furthermore, the specific process for conducting spraying experiments on the spraying equipment is as follows:

[0011] Before conducting a spraying experiment on the spraying equipment, clean the surface of the spraying equipment;

[0012] When conducting spraying experiments on the spraying equipment, drones were used to spray the equipment to simulate the actual spraying scenario; the spraying equipment included power transmission insulators, fittings, conductors, power distribution line insulators, fittings, and conductors.

[0013] Furthermore, the specific process for obtaining the improved Faster R-CNN model is as follows: obtain the Faster R-CNN model; replace the convolutional layers in the Faster R-CNN model with a ResNet structure to obtain the improved Faster R-CNN model; the improved Faster R-CNN model includes a ResNet structure and a region proposal network.

[0014] The ResNet architecture includes an extractor and a linear transformation; the extractor includes a first convolutional layer, a first pooling layer, a second pooling layer, a third pooling layer, an attention module, and a feature refinement module connected in sequence;

[0015] The region proposal network includes a foreground prediction module, which consists of a second convolutional layer, a first activation function, a third convolutional layer, and a second activation function connected in sequence.

[0016] Furthermore, the specific formula for calculating the spray coating quality fraction is as follows:

[0017] ;

[0018] In the formula, P is the spraying quality fraction; k1 represents the contribution weight of the first performance metric to the overall performance index; N represents the total area of ​​the area to be sprayed on the equipment to be sprayed. This represents the thickness of the paint to be sprayed in the i-th area to be sprayed; k1 represents the thickness of the paint that should be sprayed in the i-th area to be sprayed; k2 represents the contribution weight of the second performance metric to the overall performance index. Indicates the thickness of the coating applied by the spraying equipment; This indicates the thickness of the coating to be applied to the equipment to be coated.

[0019] Furthermore, the drone is equipped with a painting robotic arm, and the other end of the painting robotic arm is equipped with a painting device;

[0020] The painting robotic arm includes a robotic arm rotation servo, a first nozzle, a second nozzle, a nozzle telescopic device, a nozzle rotation servo, and a nozzle; the robotic arm rotation servo is mounted on the drone; one end of the first nozzle is rotatably connected to the robotic arm rotation servo, and the other end of the first nozzle is connected to one end of the second nozzle through the nozzle telescopic device. The first nozzle and the second nozzle are connected by a flexible hose, and the other end of the second nozzle is equipped with a nozzle rotation servo, which has several nozzles.

[0021] A millimeter-wave radar is installed on the side of the drone near the rotating servo motor of the robotic arm.

[0022] Furthermore, the spraying operation includes a first-mode operation, a second-mode operation, and a third-mode operation;

[0023] The first modality involves using a drone equipped with a camera to photograph the equipment to be sprayed, acquiring images of the equipment, identifying the type of equipment by using a spraying equipment type recognition model, adjusting the size and shape of the spraying robotic arm by a controller, using an edge detection algorithm to determine the area to be sprayed, and then using the spraying robotic arm to clean the surface of the area to be sprayed.

[0024] The second modal operation involves continuously adjusting the position of the spraying robotic arm and the flight path of the drone through an autonomous spraying control optimization algorithm after the surface of the area to be sprayed is cleaned, in order to complete the first round of spraying operation to identify the area to be sprayed. A camera is used to take pictures of the area to be sprayed after the first round of spraying operation to obtain a visible light image to be identified, which is then represented.

[0025] ;

[0026] In the formula, y(t) is the mass of the sprayed paint at time t; L is the system control matrix; x(t) is the flight speed of the UAV and the nozzle flow rate of the spraying robot at time t; h is the interference signal; R is the mass to be sprayed at the current time; e(t) is the system error at time t; x(t+1) is the system input at time t+1; and E is the error adjustment matrix.

[0027] The third modality involves setting a threshold for the spraying quality score after the first round of spraying is completed. The spraying quality score is obtained by identifying the visible light image to be identified through the spraying quality evaluation model. The spraying quality score is then compared with the set threshold. If the spraying quality score does not reach the set threshold, the spraying operation continues until the spraying quality score reaches the threshold.

[0028] Furthermore, the specific process of the third modality operation is as follows:

[0029] After the first round of spraying is completed, the spraying quality score of the area i to be sprayed after the first round of spraying is obtained, and matrix Q is established;

[0030] The control drone's flight speed is calculated as vi(t+1) = vi(t) * d * [1 / (Qb-Q)]; vi(t+1) represents the flight speed at time t+1 in the i-th region; vi(t) is the flight speed at time t in the i-th region; d represents the proportional coefficient of the drone's speed change; Qb represents the threshold of the evaluation score matrix.

[0031] The flow rate of the nozzle is controlled by ui(t+1) = ui(t) * a * (Qb - Q); ui(t+1) represents the nozzle flow rate at time t+1 in the i-th region; ui(t) represents the nozzle flow rate at time t in the i-th region; a represents the proportional coefficient of the change in nozzle flow rate.

[0032] The first round of spraying operations is identified by the spraying quality evaluation model. The spraying area i of the identified equipment is iteratively calculated from 0 to N until the spraying quality score of the spraying area i of the identified equipment reaches the set threshold.

[0033] A three-state operation system for spraying power transmission and distribution drones includes:

[0034] The imaging module is used to acquire visible light images captured by the vision camera after the spraying experiment on the spraying equipment, and to build a dataset based on the visible light images;

[0035] The first training module is used to pre-train the improved Faster R-CNN model using the dataset to obtain a model for identifying the types of spraying equipment.

[0036] The second training module is used to import visible light images into the convolutional neural network model for pre-training to obtain the spraying quality evaluation model.

[0037] The spraying operation module is used to acquire images of the equipment to be sprayed, identify the type of equipment by using a spraying equipment type recognition model, determine the target modal operation from multiple modal operations based on the type of equipment, perform the spraying operation on the equipment according to the target modal operation, acquire a visible light image of the equipment to be sprayed after the spraying operation, obtain a spraying quality score by using a spraying quality evaluation model to identify the visible light image, and evaluate the spraying operation using the spraying quality score.

[0038] A computer device includes: one or more processors; the processors are configured to store one or more programs; when the one or more programs are executed by the one or more processors, the three-state operation method for spraying power transmission and distribution drones is implemented.

[0039] A computer-readable storage medium having a computer program thereon, which, when executed, implements the three-state operation method for spraying power transmission and distribution drones.

[0040] Compared with existing technologies, the present invention has the following advantages:

[0041] (1) This invention provides accurate information for targeted spraying by accurately identifying the type of equipment to be sprayed and determining the area to be sprayed, avoiding misjudgment and paint waste, and improving the accuracy and reliability of the operation; in the second mode operation, the autonomous spraying control optimization algorithm dynamically adjusts the spraying parameters and corrects errors to ensure uniform spraying; in the third mode operation, a spraying quality score threshold is set, the quality is evaluated in real time and intelligent decisions are made on whether to continue the operation to ensure that the spraying quality meets the requirements; in addition, this method can identify a variety of power transmission and distribution equipment, adapt to different types of characteristics for spraying, and UAV operation can overcome the limitations of complex environment, improve the safety and efficiency of operation, and has strong versatility and adaptability.

[0042] (2) This invention establishes a matrix Q to quantitatively evaluate the spraying quality of each area through a three-modal operation method. Based on the spraying quality score, the flight speed of the UAV and the nozzle flow rate are dynamically adjusted to achieve precise quality control and intelligent parameter adjustment. At the same time, the spraying quality evaluation model is used to iteratively calculate until the spraying quality score of each area meets the standard, ensuring full spraying coverage, thereby improving the reliability, spraying efficiency and quality stability of the spraying equipment and extending the service life of the equipment. Attached Figure Description

[0043] Figure 1 This is a flowchart of the method of the present invention.

[0044] Figure 2 This is a structural diagram of the extractor of the present invention.

[0045] Figure 3 This is a structural diagram of the foreground judgment module of the present invention.

[0046] Figure 4 This is a schematic diagram of the drone and the painting robotic arm of the present invention.

[0047] The markings in the diagram are: 1. Unmanned aerial vehicle (UAV); 2. Robotic arm rotation servo motor; 3. First nozzle; 31. Second nozzle; 4. Nozzle telescopic device; 5. Nozzle rotation servo motor; 6. Nozzle; 7. Millimeter-wave radar; 8. Spraying equipment. Detailed Implementation

[0048] like Figure 1 As shown, the present invention provides a technical solution: a three-state operation method for spraying power transmission and distribution drones, comprising the following steps:

[0049] Step S1: Acquire visible light images captured by a vision camera after the spraying experiment on the spraying equipment, and construct a dataset based on the visible light images;

[0050] Step S2: Use the dataset to pre-train the improved Faster R-CNN model to obtain a spraying equipment type recognition model;

[0051] Step S3: Import the visible light image into the convolutional neural network model for pre-training to obtain the spraying quality evaluation model;

[0052] Step S4: Acquire an image of the equipment to be sprayed, identify the type of equipment by using a spraying equipment type recognition model, determine the target modal operation from multiple modal operations based on the type of equipment to be sprayed, perform spraying operation on the equipment to be sprayed according to the target modal operation, acquire a visible light image to be identified after the spraying operation on the equipment to be sprayed, obtain a spraying quality score by using a spraying quality evaluation model to identify the visible light image to be identified, and use the spraying quality score to evaluate the spraying operation.

[0053] The specific process for obtaining the spraying quality evaluation model is as follows:

[0054] Visible light images are imported into a convolutional neural network model for pre-training; the pre-trained convolutional neural network model is obtained; the dataset is augmented to obtain augmented samples, which are then imported into the pre-trained convolutional neural network model. The SGDM algorithm is used to minimize the cross-entropy loss function to optimize the pre-trained convolutional neural network model, resulting in a spraying quality evaluation model.

[0055] The specific process of constructing the dataset is as follows:

[0056] Visible light images of the spraying equipment were collected, including images taken under different weather conditions, environments, and shooting angles to enhance sample diversity. These images encompassed the operational environments of power distribution line insulators in various real-world scenarios. The visible light images were labeled and categorized, and a dataset was constructed based on these labeled and categorized images. The spraying equipment included transmission line insulators, fittings, conductors, and distribution line insulators, fittings, and conductors. For distribution line insulators, the dataset covered voltage levels of 35kV and above, including various types of towers such as straight-line towers and tension towers.

[0057] The specific process of conducting spraying experiments on the spraying equipment is as follows:

[0058] Before conducting a spraying experiment on the spraying equipment, clean the surface of the spraying equipment to ensure that the paint can adhere fully.

[0059] When conducting spraying experiments on the equipment to be sprayed, drones were used to spray the equipment to simulate the actual spraying scenario. The equipment to be sprayed includes six categories of equipment: power transmission insulators, fittings, conductors, distribution line insulators, fittings and conductors.

[0060] The specific process for obtaining the improved Faster R-CNN model is as follows:

[0061] Obtain the Faster R-CNN model; replace the convolutional layers in the Faster R-CNN model with a ResNet structure to obtain the improved Faster R-CNN model; the improved Faster R-CNN model includes a ResNet structure and a Region Proposal Network (RPN).

[0062] like Figure 2 As shown; the ResNet structure includes an extractor and a linear transformation; the extractor includes a first convolutional layer (2×2), a first pooling layer (2×2), a second pooling layer (2×2), a third pooling layer (2×2), an attention module (ResNeSt), and a feature refinement block; after the first convolutional layer, the first pooling layer, the second pooling layer, the third pooling layer, the attention module, and the feature refinement block are added sequentially;

[0063] The visible light image is processed sequentially through the first convolutional layer, the first pooling layer, the second pooling layer, the third pooling layer, and the attention module in the extractor. The output of the attention module is then input into the feature refinement module for further processing, yielding the output of the feature refinement module, which is the feature map. The extractor using the ResNet structure extracts features from the visible light image, obtaining the feature map. The extractor further compresses the size of the visible light image while improving the extraction of key features.

[0064] The linear transformation is a fully connected layer used to map the input visible light image to the output vector. To reduce the computational cost of the improved Faster R-CNN model, the linear transformation is optimized to reduce the number of fully connected layers. In addition, batch regularization is used to normalize the output vector mapped from the visible light image, making the training of the improved Faster R-CNN model easier.

[0065] like Figure 3 As shown; the region proposal network includes a foreground detection module, which includes a second convolutional layer (3×3), a first activation function, a third convolutional layer (1×1), and a second activation function; the first activation function, the third convolutional layer, and the second activation function are added sequentially after the second convolutional layer;

[0066] The role of the region proposal network is to generate candidate boxes for feature map selection. An improvement to the original region proposal network is made: instead of generating a fixed number of prior boxes, the feature map is first convolved, and prior boxes are generated proportionally based on the number of channels extracted from the feature map, reducing unnecessary operations. In the foreground and background judgment module, edge features exist in the boundary region between the foreground and background. Therefore, the optimized method for the foreground judgment module is to pass through a second convolutional layer. If the first activation function cannot judge the feature map, a third convolutional layer (1×1) is performed to refine the feature map. Subsequently, the second activation function outputs the foreground and background categories of the feature map. If the confidence score of either the foreground or background category exceeds a 50% threshold set for that category, it is considered that category, which is the equipment to be sprayed.

[0067] The specific formula for calculating the spray coating quality fraction is as follows:

[0068] ;

[0069] In the formula, P is the spraying quality fraction; k1 represents the contribution weight of the first performance metric to the overall performance index; N represents the total area of ​​the area to be sprayed on the equipment to be sprayed. This represents the thickness of the paint to be sprayed in the i-th area to be sprayed; k1 represents the thickness of the paint that should be sprayed in the i-th area to be sprayed; k2 represents the contribution weight of the second performance metric to the overall performance index. Indicates the thickness of the coating applied by the spraying equipment; This indicates the thickness of the coating to be applied to the equipment to be coated.

[0070] The data set enhancement process includes geometric transformation, pixel transformation, and sample blending. Geometric transformation methods include image flipping, rotation, scaling, translation, and dithering. Pixel transformation involves adjusting the pixel values ​​of visible light images, adding salt-and-pepper noise, Gaussian noise, applying Gaussian blur, and adjusting HSV contrast, brightness, saturation, histogram equalization, and white balance, which can change the color, brightness, and texture features of visible light images.

[0071] like Figure 4 As shown; wherein, the drone 1 is equipped with a spraying robotic arm, and the other end of the spraying robotic arm is equipped with a spraying device 8;

[0072] The painting robotic arm includes a robotic arm rotation servo 2, a first nozzle 3, a second nozzle 31, a nozzle telescopic device 4, a nozzle rotation servo 5, and a nozzle 6. The robotic arm rotation servo 2 is mounted on the UAV 1. One end of the first nozzle 3 is rotatably connected to the robotic arm rotation servo 2, and the other end of the first nozzle 3 is connected to one end of the second nozzle 31 through the nozzle telescopic device 4. The first nozzle 3 and the second nozzle 31 are connected by a hose. The other end of the second nozzle 31 is equipped with a nozzle rotation servo 5, and the nozzle rotation servo 5 is equipped with several nozzles 6.

[0073] Among them, a millimeter-wave radar 7 is installed on the side of the UAV 1 near the rotating servo motor 2 of the robotic arm.

[0074] The drone is equipped with a controller, which is the central CPU of the drone and is used to adjust the size and shape of the spraying robotic arm.

[0075] The specific control is as follows: the robotic arm rotation servo motor 2 allows for arbitrary adjustment of the spraying robotic arm's horizontal to vertical angle. When spraying insulators, or during the takeoff and landing of the UAV 1, the robotic arm rotation servo motor 2 rotates the spraying robotic arm to a horizontal position. When spraying wires, it can be rotated to a vertical position for convenient wire spraying. The first spray nozzle 3 and the second spray nozzle 31 are made of high-strength epoxy resin, possessing high mechanical strength and providing insulation. The spray nozzle extension device 4 can adjust the length of the second spray nozzle 31 to adapt to different voltage levels. Different voltages require different insulation lengths, thus allowing for flexible adjustment when applied to power transmission or distribution equipment. The nozzle rotation servo motor 5 can adjust the shape of the nozzle 6, making it suitable for different equipment. When spraying insulators, the nozzle 6 can be adjusted to a U-shape or triangle to simultaneously spray multiple surfaces of the insulator. When spraying hardware, it can be adjusted to a straight line to increase the spray coverage area and improve work efficiency.

[0076] The spraying operation includes the first modal operation, the second modal operation, and the third modal operation;

[0077] The first modality involves using a drone equipped with a camera to photograph the equipment to be sprayed, acquiring images of the equipment, identifying the type of equipment by using a spraying equipment type recognition model, adjusting the size and shape of the spraying robotic arm by a controller, using an edge detection algorithm to determine the area to be sprayed, and then using the spraying robotic arm to clean the surface of the area to be sprayed.

[0078] The second modal operation involves cleaning the surface of the area to be sprayed on the equipment, and then continuously adjusting the position of the spraying robot arm and the flight path of the drone through an autonomous spraying control optimization algorithm to complete the first round of spraying operation to identify the area to be sprayed. A camera is used to take pictures of the area to be sprayed after the first round of spraying operation to obtain a visible light image to be identified, which is then represented.

[0079] ;

[0080] In the formula, y(t) is the mass of the sprayed paint at time t; L is the system control matrix; x(t) is the flight speed of the UAV and the nozzle flow rate of the spraying robot at time t; h is the interference signal; R is the mass to be sprayed at the current time; e(t) is the system error at time t; x(t+1) is the system input at time t+1; and E is the error adjustment matrix.

[0081] The third modality involves setting a threshold for the spraying quality score after the first round of spraying is completed. The spraying quality evaluation model is used to identify the visible light image to obtain the spraying quality score. The spraying quality score is then compared with the set threshold. If the spraying quality score does not reach the set threshold, the spraying operation continues until the spraying quality score reaches the threshold.

[0082] The specific process of the third modality operation is as follows:

[0083] After the first round of spraying is completed, the spraying quality score of the area i to be sprayed after the first round of spraying is obtained, and matrix Q is established;

[0084] Step S52: Control the flight speed of the UAV vi(t+1) = vi(t) * d * [1 / (Qb-Q)]; vi(t+1) represents the flight speed of the i-th region at time t+1; vi(t) is the flight speed of the i-th region at time t; d represents the proportional coefficient of the UAV speed change; Qb represents the threshold of the evaluation score matrix;

[0085] The flow rate of the nozzle is controlled by ui(t+1) = ui(t) * a * (Qb - Q); ui(t+1) represents the nozzle flow rate at time t+1 in the i-th region; ui(t) represents the nozzle flow rate at time t in the i-th region; a represents the proportional coefficient of the change in nozzle flow rate.

[0086] The first round of spraying operations is identified by the spraying quality evaluation model. The spraying area i of the identified equipment is iteratively calculated from 0 to N until the spraying quality score of the spraying area i of the identified equipment reaches the set threshold.

[0087] A three-state operation system for spraying power transmission and distribution drones includes:

[0088] The imaging module is used to acquire visible light images captured by the vision camera after the spraying experiment on the spraying equipment, and to build a dataset based on the visible light images;

[0089] The first training module is used to pre-train the improved Faster R-CNN model using the dataset to obtain a model for identifying the types of spraying equipment.

[0090] The second training module is used to import visible light images into the convolutional neural network model for pre-training to obtain the spraying quality evaluation model.

[0091] The spraying operation module is used to acquire images of the equipment to be sprayed, identify the type of equipment by using a spraying equipment type recognition model, determine the target modal operation from multiple modal operations based on the type of equipment, perform the spraying operation on the equipment according to the target modal operation, acquire a visible light image of the equipment to be sprayed after the spraying operation, obtain a spraying quality score by using a spraying quality evaluation model to identify the visible light image, and evaluate the spraying operation using the spraying quality score.

[0092] A computer device includes: one or more processors; the processors are configured to store one or more programs; when the one or more programs are executed by the one or more processors, the three-state operation method for spraying power transmission and distribution drones is implemented.

[0093] A computer-readable storage medium having a computer program thereon, which, when executed, implements the three-state operation method for spraying power transmission and distribution drones.

[0094] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0095] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0096] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0100] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A three-state operation method for spraying paint on power transmission and distribution drones, characterized in that, Includes the following steps: Acquire visible light images captured by a vision camera after a spraying experiment on a spraying equipment, and construct a dataset based on the visible light images; The improved Faster R-CNN model was pre-trained using the dataset to obtain a model for identifying the types of spraying equipment. Visible light images are imported into a convolutional neural network model for pre-training to obtain a coating quality evaluation model. The process involves acquiring an image of the equipment to be sprayed, identifying the type of equipment using a spraying equipment type recognition model, determining the target modal operation from multiple modal operations based on the type of equipment, performing a spraying operation on the equipment according to the target modal operation, acquiring a visible light image of the equipment to be sprayed after the spraying operation, obtaining a spraying quality score by identifying the visible light image using a spraying quality evaluation model, and evaluating the spraying operation using the spraying quality score. The spraying operation includes a first-mode operation, a second-mode operation, and a third-mode operation; The first modality involves using a drone equipped with a camera to photograph the equipment to be sprayed, acquiring images of the equipment, identifying the type of equipment by using a spraying equipment type recognition model, adjusting the size and shape of the spraying robotic arm by a controller, using an edge detection algorithm to determine the area to be sprayed, and then using the spraying robotic arm to clean the surface of the area to be sprayed. The second modal operation involves continuously adjusting the position of the spraying robotic arm and the flight path of the drone through an autonomous spraying control optimization algorithm after the surface of the area to be sprayed is cleaned, in order to complete the first round of spraying operation to identify the area to be sprayed. A camera is used to take pictures of the area to be sprayed after the first round of spraying operation to obtain a visible light image to be identified, which is then represented. ； In the formula, y(t) is the mass of the sprayed paint at time t; L is the system control matrix; x(t) is the flight speed of the UAV and the nozzle flow rate of the spraying robot at time t; h is the interference signal; R is the mass to be sprayed at the current time; e(t) is the system error at time t; x(t+1) is the system input at time t+1; and E is the error adjustment matrix. The third modal operation involves setting a threshold for the spraying quality score after the first round of spraying is completed. The spraying quality score is obtained by identifying the visible light image to be identified through the spraying quality evaluation model. The spraying quality score is then compared with the set threshold. If the spraying quality score does not reach the set threshold, the spraying operation continues until the spraying quality score reaches the threshold. The specific process of the third modal operation is as follows: After the first round of spraying is completed, the spraying quality score of the area i to be sprayed after the first round of spraying is obtained, and matrix Q is established; The control drone's flight speed is calculated as vi(t+1) = vi(t) * d * [1 / (Qb-Q)]; vi(t+1) represents the flight speed at time t+1 in the i-th region; vi(t) is the flight speed at time t in the i-th region; d represents the proportional coefficient of the drone's speed change; Qb represents the threshold of the evaluation score matrix. The flow rate of the nozzle is controlled by ui(t+1) = ui(t) * a * (Qb - Q); ui(t+1) represents the nozzle flow rate at time t+1 in the i-th region; ui(t) represents the nozzle flow rate at time t in the i-th region; a represents the proportional coefficient of the change in nozzle flow rate. The first round of spraying operations is identified by the spraying quality evaluation model. The spraying area i of the identified equipment is iteratively calculated from 0 to N until the spraying quality score of the spraying area i of the identified equipment reaches the set threshold.

2. The three-state operation method for spraying power transmission and distribution UAVs according to claim 1, characterized in that: The specific process for conducting a spraying experiment on the spraying equipment is as follows: Before conducting a spraying experiment on the spraying equipment, clean the surface of the spraying equipment; When conducting spraying experiments on the spraying equipment, drones were used to spray the equipment to simulate the actual spraying scenario; the spraying equipment included power transmission insulators, fittings, conductors, power distribution line insulators, fittings, and conductors.

3. The three-state operation method for spraying power transmission and distribution UAVs according to claim 2, characterized in that: The specific process for obtaining the improved Faster R-CNN model is as follows: obtain the Faster R-CNN model; replace the convolutional layers in the Faster R-CNN model with a ResNet structure to obtain the improved Faster R-CNN model; the improved Faster R-CNN model includes a ResNet structure and a region proposal network. The ResNet architecture includes an extractor and a linear transformation; the extractor includes a first convolutional layer, a first pooling layer, a second pooling layer, a third pooling layer, an attention module, and a feature refinement module connected in sequence; The region proposal network includes a foreground prediction module, which consists of a second convolutional layer, a first activation function, a third convolutional layer, and a second activation function connected in sequence.

4. The three-state operation method for spraying power transmission and distribution UAVs according to claim 1, characterized in that: The specific formula for calculating the spray coating quality fraction is as follows: ； In the formula, P is the spraying quality fraction; k1 represents the contribution weight of the first performance metric to the overall performance index; N represents the total area of ​​the area to be sprayed on the equipment to be sprayed. This represents the thickness of the paint to be sprayed in the i-th area to be sprayed; k1 represents the thickness of the paint that should be sprayed in the i-th area to be sprayed; k2 represents the contribution weight of the second performance metric to the overall performance index. Indicates the thickness of the coating applied by the spraying equipment; This indicates the thickness of the coating to be applied to the equipment to be coated.

5. The three-state operation method for spraying power transmission and distribution UAVs according to claim 2, characterized in that: The drone is equipped with a painting robotic arm, and the other end of the painting robotic arm is equipped with a painting device; The painting robotic arm includes a robotic arm rotation servo, a first nozzle, a second nozzle, a nozzle telescopic device, a nozzle rotation servo, and a nozzle; the robotic arm rotation servo is mounted on the drone; one end of the first nozzle is rotatably connected to the robotic arm rotation servo, and the other end of the first nozzle is connected to one end of the second nozzle through the nozzle telescopic device. The first nozzle and the second nozzle are connected by a flexible hose, and the other end of the second nozzle is equipped with a nozzle rotation servo, which has several nozzles. A millimeter-wave radar is installed on the side of the drone near the rotating servo motor of the robotic arm.

6. A three-state operation system for spraying paint on a power transmission and distribution drone, used to execute the three-state operation method for spraying paint on a power transmission and distribution drone as described in any one of claims 1 to 5, characterized in that, include: The imaging module is used to acquire visible light images captured by the vision camera after the spraying experiment on the spraying equipment, and to build a dataset based on the visible light images; The first training module is used to pre-train the improved Faster R-CNN model using the dataset to obtain a spraying equipment type recognition model. The second training module is used to import visible light images into the convolutional neural network model for pre-training to obtain the spraying quality evaluation model. The spraying operation module is used to acquire images of the equipment to be sprayed, identify the type of equipment by using a spraying equipment type recognition model, determine the target modal operation from multiple modal operations based on the type of equipment, perform the spraying operation on the equipment according to the target modal operation, acquire a visible light image of the equipment to be sprayed after the spraying operation, obtain a spraying quality score by using a spraying quality evaluation model to identify the visible light image, and evaluate the spraying operation using the spraying quality score.

7. A computer device, characterized in that, include: One or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a three-state operation method for spraying power transmission and distribution UAVs as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements a three-state operation method for spraying power transmission and distribution drones as described in any one of claims 1 to 5.

Images