Method and system for detecting defects of power transmission line under multiple working conditions

Abstract

The invention relates to the technical field of power transmission lines, and discloses a method and a system for detecting defects of a power transmission line under multiple working conditions. The detection method comprises the following steps: acquiring an initial image book of a power transmission line by adopting inspection equipment, and preprocessing the initial image book to generate a basic image training book; adopting the basic image training book to train a model platform built by a YOLOv5 algorithm to obtain a first detection model; selecting a proper loss function to optimize parameters of the first detection model, and adjusting to obtain a second detection model; and deploying the second detection model in inspection equipment for quickly positioning and identifying defects in the power transmission line. The detection method improves the identification efficiency and accuracy of the defects of the power transmission line in the inspection process, and is high in applicability and timeliness.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission lines, and in particular to a method and system for detecting power transmission line defects under multiple working conditions. Background Art

[0002] As transmission lines serve as important channels for power transmission, the stability and safety of their operating status are of vital importance. However, since there are so many transmission lines and they are usually exposed to complex and changeable environments, such as high temperature, severe cold, strong winds, heavy rain, etc., these working conditions will cause varying degrees of damage and defects to the transmission lines, resulting in increased losses and lower efficiency of power grid equipment, and even major accidents such as equipment damage, affecting people’s lives and property safety. Traditional methods for detecting defects in transmission lines mainly rely on manual inspections and semi-automatic instruments, resulting in low work efficiency and low recognition accuracy. Some existing inspection equipment methods use drones to store the images to be detected in local files first, and then manually import them to the server after returning to the office. This has low timeliness and cannot be used for real-time monitoring and positioning. Summary of the invention

[0003] In order to overcome the above problems, the present invention provides a method and system for detecting defects in power transmission lines under multiple working conditions. The method uses patrol equipment to collect initial image albums of power transmission lines, and preprocesses the initial image albums to generate basic image training albums. At the same time, the basic image training albums are used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model; a suitable loss function is selected to optimize the parameters of the first detection model, and a second detection model is adjusted to obtain the second detection model. The second detection model is deployed in the patrol equipment for quickly locating and identifying defects in the power transmission line, thereby improving the efficiency and accuracy of identifying defects in the power transmission line during the patrol process. At the same time, the method has strong applicability and high timeliness.

[0004] In order to achieve the above object, the present invention provides a method for detecting defects in power transmission lines under multiple working conditions, characterized in that the detection method comprises:

[0005] Using inspection equipment to collect an initial image album of the transmission line, and preprocessing the initial image album to generate a basic image training album;

[0006] The basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model;

[0007] Selecting a suitable loss function to optimize the parameters of the first detection model, and adjusting to obtain a second detection model;

[0008] The second detection model is deployed in the inspection equipment to quickly locate and identify defects in the transmission line.

[0009] Preferably, an initial image album of the power transmission line is collected by using inspection equipment, and the initial image album is preprocessed to generate a basic image training album, including:

[0010] Using inspection equipment to collect initial images of the transmission line under different working conditions;

[0011] marking defects in the initial image album;

[0012] The initial image albums of different working conditions are subjected to image preprocessing to generate the basic image training albums.

[0013] Preferably, the initial image album of the power transmission line under different working conditions is collected by inspection equipment, including:

[0014] Different imaging techniques are used to acquire visible light images, infrared light images and ultraviolet light images of the transmission line under different working conditions to form the initial image album.

[0015] Preferably, image preprocessing is performed on the initial image albums of different working conditions to generate the basic image training album, including:

[0016] Performing image denoising and enhancement processing on the initial image album;

[0017] Performing blurring and noise adding processing on the image of the initial image album containing the defect;

[0018] The data enhancement technology is used to rotate and scale the initial image album to form the basic image training album.

[0019] Preferably, the basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model, including:

[0020] An input terminal is provided in the input network of the model platform for receiving and processing the base

[0021] In the Backbone network of the model platform, a convolution layer, batch normalization and activation function are provided as feature extractors;

[0022] Setting FPN and PAN network structures in the neck network of the model platform to achieve multi-scale fusion of feature maps;

[0023] An anchor frame and a prediction layer are set in the head network of the model platform to provide detection of the feature map to form the first detection model.

[0024] Preferably, selecting a suitable loss function to optimize the parameters of the first detection model and adjusting to obtain the second detection model includes:

[0025] Choose an appropriate loss function;

[0026] Initializing the first detection model and performing iterative training using an optimizer;

[0027] regularly obtaining a loss function of the first detection model and a reliability index parameter of the first detection model, and adjusting the parameters of the first detection model;

[0028] Calculate the gradient using a back propagation algorithm based on the loss function;

[0029] Using a stochastic gradient descent (SGD) algorithm to update the parameters of the first detection model based on the gradient;

[0030] When the loss value of the loss function and the weighted value of the reliability indicator parameter reach a preset threshold, the current detection model and its parameters are maintained to obtain the second detection model.

[0031] Preferably, a suitable loss function is selected, including:

[0032] Based on the bounding box regression loss and confidence loss of the YOLOv5 algorithm, a mean square error loss function and a cross entropy loss function are selected respectively, and a weighted sum loss function of the mean square error loss function and the cross entropy loss function is defined as the loss function of the first detection model.

[0033] Preferably, the reliability indicator parameters include accuracy and recall.

[0034] Preferably, the second detection model is deployed in the inspection device to quickly locate and identify defects in the transmission line, including:

[0035] Using a drone equipped with an image acquisition device to inspect the transmission line;

[0036] The drone monitors and photographs the power transmission line to generate an image to be detected;

[0037] The image is input into the second detection model to identify and locate defects of the transmission line.

[0038] On the other hand, the present invention provides a detection system for power transmission line defects under multiple working conditions, comprising:

[0039] A data acquisition and processing module, which acquires and processes image data of the power transmission line under different working conditions to generate a basic image training book;

[0040] The second detection model module is used to install the second detection model, is electrically connected to the data acquisition and processing module, and is used to execute any of the detection methods described above.

[0041] Through the above technical scheme, the method uses patrol equipment to collect the initial image album of the transmission line, including visible light images, infrared images and ultraviolet images, and pre-processes the initial image album, marks defects and classifies different working conditions, and uses image enhancement, data amplification and other technologies to generate a rich image set to generate a basic image training album. At the same time, the basic image training album is used to train the model platform built by the YOLOv5 algorithm to obtain the first detection model, reasonably configure the network structure and parameters, realize the extraction and fusion of feature information of the image, improve the robustness of the monitoring model, select a suitable loss function to optimize the parameters of the first detection model, adjust and obtain the second detection model, ensure the applicability and practicality of the detection model, and deploy the second detection model in the patrol equipment for rapid positioning and identification of defects in the transmission line, improve the efficiency and accuracy of identifying defects in the transmission line during the patrol inspection, and at the same time, have strong applicability and high timeliness. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0043] Figure 1 is a flow chart of a method for detecting power transmission line defects under multiple working conditions according to an embodiment of the present invention;

[0044] Figure 2 is a flow chart of generating a first detection model for a method for detecting defects of a power transmission line under multiple working conditions according to an embodiment of the present invention;

[0045] Figure 3 It is a flowchart of generating a second detection model for a method for detecting defects in a power transmission line under multiple working conditions according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0047] like Figure 1 FIG. 1 is a flow chart of a method for detecting power transmission line defects under multiple working conditions according to an embodiment of the present invention. Figure 1 In the present invention, the detection method may include:

[0048] In step S10, an initial image album of the power transmission line is collected by using inspection equipment, and the initial image album is preprocessed to generate a basic image training album;

[0049] In step S11, the basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model;

[0050] In step S12, a suitable loss function is selected to optimize the parameters of the first detection model, and the second detection model is adjusted to obtain;

[0051] In step S13, the second detection model is deployed in the inspection equipment to quickly locate and identify defects in the transmission line.

[0052] In such Figure 1 In the method shown, step S10 uses inspection equipment to collect an initial image album of the transmission line, and preprocesses the initial image album to generate a basic image training album; obtains initial image albums under different working conditions, the initial image albums include visible light images, infrared light images and ultraviolet light images, removes redundant images in the original image set, and classifies and preprocesses the images to ensure the quality and accuracy of the images.

[0053] Step S11 uses the basic image training book to train the model platform built by the YOLOv5 algorithm to obtain a first detection model, and reasonably configures the network structure and parameters. The model can be used for extracting feature information and multi-scale fusion of feature maps to improve detection efficiency.

[0054] Step S12 selects a suitable loss function to optimize the parameters of the first detection model, adjusts and obtains the second detection model, selects a suitable loss function for weighted processing for the bounding box regression loss and confidence loss of the YOLOv5 algorithm, and obtains the gradient to update the parameters according to the stochastic gradient descent SGD algorithm to obtain the second detection model;

[0055] Step S13 deploys the second detection model in the inspection equipment for quickly locating and identifying defects in the transmission line, and uses inspection equipment such as drones to carry the second detection model for real-time monitoring to improve timeliness and work efficiency.

[0056] The method uses patrol equipment to collect initial image albums of transmission lines, preprocesses the initial image albums, generates basic image training albums, and uses the basic image training albums to train the model platform built by the YOLOv5 algorithm to obtain a first detection model; selects a suitable loss function to optimize the parameters of the first detection model, adjusts and obtains the second detection model, and deploys the second detection model in the patrol equipment to quickly locate and identify defects in the transmission line, thereby improving the efficiency and accuracy of identifying defects in the transmission line during the inspection process, and at the same time has strong applicability and high timeliness.

[0057] In order to improve the efficiency and accuracy of transmission line detection and ensure the location of defects on the transmission line, in one embodiment of the present invention, the initial image album of the transmission line is collected by patrol equipment, and the initial image album is preprocessed to generate a basic image training album, which may include the following steps: using patrol equipment to collect the initial image album of the transmission line under different working conditions; marking the defects in the initial image album; preprocessing the initial image album of different working conditions to generate a basic image training album, so that the basic image training album can be used to train the detection model. At the same time, based on the need to meet the transmission line imaging under different working conditions in the image of the initial image album, in one embodiment of the present invention, the initial image album of the transmission line under different working conditions is collected by patrol equipment, including using different image technologies to obtain the visible light image, infrared light image and ultraviolet light image of the transmission line under different working conditions to form an initial image album, ensuring that more abundant image information of the transmission line can be provided and improving processing efficiency.

[0058] In order to enrich the image information and improve the generalization ability of the model, in one embodiment of the present invention, the initial image albums of different working conditions are subjected to image preprocessing to generate a basic image training album, including the following steps: performing image denoising and enhancement processing on the initial image album; performing blurring and noise addition processing on the images in the initial image album containing defects; and rotating and scaling the initial image album using data enhancement technology to form a basic image training album, thereby laying a foundation for training the detection model and improving the detection efficiency.

[0059] Figure 2 The present invention is a flowchart of generating a first detection model of a method for detecting defects in power transmission lines under multiple working conditions according to an embodiment of the present invention; considering the responsiveness of the detection model, the network structure of the detection model is reasonably configured. In an embodiment of the present invention, the basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain the first detection model, including the following steps:

[0060] Step S20: an input terminal is provided in the input network of the model platform for receiving and processing image information of the basic image training book;

[0061] Step S21 sets a convolution layer, batch normalization and activation function as a feature extractor in the Backbone network of the model platform;

[0062] Step S22 sets the FPN and PAN network structures in the neck network of the model platform to achieve multi-scale fusion of feature maps;

[0063] Step S23 sets an anchor frame and a prediction layer in the head network of the model platform to provide feature map detection to form the first detection model.

[0064] Figure 3 This is a flowchart of generating a second detection model for a method for detecting defects in power transmission lines under multiple working conditions according to an embodiment of the present invention. In order to ensure the reliability of the first detection model, a reasonable detection model is optimized and adjusted; in one embodiment of the present invention, a suitable loss function is selected to optimize the parameters of the first detection model, and the second detection model is adjusted and obtained, including the following steps:

[0065] Step S30: selecting a suitable loss function;

[0066] Step S31 initializes the first detection model and uses an optimizer to perform iterative training;

[0067] Step S32: regularly obtaining the loss function of the first detection model and the reliability index parameters of the first detection model, and adjusting the parameters of the first detection model;

[0068] Step S33 calculates the gradient using a back propagation algorithm based on the loss function;

[0069] Step S34 uses the stochastic gradient descent SGD algorithm to update the parameters of the first detection model based on the gradient;

[0070] In step S35, when the loss value of the loss function and the weighted value of the reliability index parameter reach a preset threshold, the current detection model and its parameters are maintained to obtain a second detection model.

[0071] Taking into account the needs based on the YOLOv5 algorithm, in order to improve the reliability and detection performance of the second detection model, in one embodiment of the present invention, selecting a suitable loss function can be based on the bounding box regression loss and confidence loss of the YOLOv5 algorithm, and selecting a mean square error loss function and a cross entropy loss function respectively, and defining the weighted sum loss function of the mean square error loss function and the cross entropy loss function as the loss function of the first detection model; the reliability indicator parameters may include accuracy and recall.

[0072] In order to improve the detection efficiency of the inspection equipment, in one embodiment of the present invention, the second detection model is deployed in the inspection equipment to quickly locate and identify defects in the transmission line, including the following steps:

[0073] Use drones equipped with image acquisition equipment to inspect power transmission lines;

[0074] Drones monitor and photograph transmission lines to generate images to be inspected;

[0075] The image is input into the second detection model to realize the identification and location of the defects of the transmission line.

[0076] A second aspect of the present invention provides a detection system for transmission line defects under multiple working conditions. The detection system includes a data acquisition and processing module and a second detection model module. The data acquisition and processing module acquires and processes image data of the transmission line under different working conditions to generate a basic image training book. The second detection model module is used to install a second detection model and is electrically connected to the data acquisition and processing module to execute any of the detection methods described above.

[0077] Through the above technical scheme, the method uses patrol equipment to collect the initial image album of the transmission line, including visible light images, infrared images and ultraviolet images, and pre-processes the initial image album, marks defects and classifies different working conditions, and uses image enhancement, data amplification and other technologies to generate a rich image set to generate a basic image training album. At the same time, the basic image training album is used to train the model platform built by the YOLOv5 algorithm to obtain the first detection model, reasonably configure the network structure and parameters, realize the extraction and fusion of feature information of the image, improve the robustness of the monitoring model, select a suitable loss function to optimize the parameters of the first detection model, adjust and obtain the second detection model, ensure the applicability and practicality of the detection model, and deploy the second detection model in the patrol equipment for rapid positioning and identification of defects in the transmission line, improve the efficiency and accuracy of identifying defects in the transmission line during the patrol inspection, and at the same time, have strong applicability and high timeliness.

[0078] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0079] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting defects in power transmission lines under multiple working conditions, characterized in that: The detection method comprises: Using inspection equipment to collect an initial image album of the transmission line, and preprocessing the initial image album to generate a basic image training album; The basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model; Selecting a suitable loss function to optimize the parameters of the first detection model, and adjusting to obtain a second detection model; The second detection model is deployed in the inspection equipment to quickly locate and identify defects in the transmission line.

2. The detection method according to claim 1, characterized in that: An initial image album of the transmission line is collected by inspection equipment, and the initial image album is preprocessed to generate a basic image training album, including: Using inspection equipment to collect initial images of the transmission line under different working conditions; marking defects in the initial image album; The initial image albums of different working conditions are subjected to image preprocessing to generate the basic image training albums.

3. The detection method according to claim 2, characterized in that: The inspection equipment is used to collect initial images of the transmission line under different working conditions, including: Different imaging techniques are used to acquire visible light images, infrared light images and ultraviolet light images of the transmission line under different working conditions to form the initial image album.

4. The detection method according to claim 2, characterized in that: Performing image preprocessing on the initial image albums of different working conditions to generate the basic image training albums includes: Performing image denoising and enhancement processing on the initial image album; Performing blurring and noise adding processing on the image of the initial image album containing the defect; The data enhancement technology is used to rotate and scale the initial image album to form the basic image training album.

5. The detection method according to claim 1, characterized in that: The basic image training book is used to train the model platform built by the YOLOv5 algorithm to obtain a first detection model, including: An input terminal is provided in the input network of the model platform for receiving and processing image information of the basic image training book; In the Backbone network of the model platform, a convolution layer, batch normalization and activation function are provided as feature extractors; Setting FPN and PAN network structures in the neck network of the model platform to achieve multi-scale fusion of feature maps; An anchor frame and a prediction layer are set in the head network of the model platform to provide detection of the feature map to form the first detection model.

6. The detection method according to claim 5, characterized in that: Selecting a suitable loss function to optimize the parameters of the first detection model and adjusting to obtain the second detection model includes: Choose an appropriate loss function; Initializing the first detection model and performing iterative training using an optimizer; regularly obtaining a loss function of the first detection model and a reliability index parameter of the first detection model, and adjusting the parameters of the first detection model; Calculate the gradient using a back propagation algorithm based on the loss function; Using a stochastic gradient descent (SGD) algorithm to update the parameters of the first detection model based on the gradient; When the loss value of the loss function and the weighted value of the reliability indicator parameter reach a preset threshold, the current detection model and its parameters are maintained to obtain the second detection model.

7. The detection method according to claim 6, characterized in that: Choose an appropriate loss function, including: Based on the bounding box regression loss and confidence loss of the YOLOv5 algorithm, a mean square error loss function and a cross entropy loss function are selected respectively, and a weighted sum loss function of the mean square error loss function and the cross entropy loss function is defined as the loss function of the first detection model.

8. The detection method according to claim 6, characterized in that: The reliability index parameters include accuracy and recall.

9. The detection method according to claim 1, characterized in that: Deploying the second detection model in the inspection device to quickly locate and identify defects in the transmission line includes: Using a drone equipped with an image acquisition device to inspect the transmission line; The drone monitors and photographs the power transmission line to generate an image to be detected; The image is input into the second detection model to realize identification and location of defects of the transmission line.

10. A transmission line defect detection system for multiple working conditions, characterized in that: include: A data acquisition and processing module, which acquires and processes image data of the power transmission line under different working conditions to generate a basic image training book; The second detection model module is used to install the second detection model, is electrically connected to the data acquisition and processing module, and is used to execute the detection method as described in any one of claims 1-9.

Images