Power transmission line intelligent inspection method based on image perception

Abstract

A power transmission line intelligent inspection method based on image perception relates to the field of power transmission line inspection. The problem that when an existing traditional image processing method is applied to a power transmission line inspection task, due to the fact that the detection and segmentation mode is greatly influenced by the image background, brightness distribution and other factors, the traditional method lacks a large amount of priori knowledge and cannot deal with a complex and changeable scene environment is solved. The method comprises the following steps: acquiring images of power transmission line components under different meteorological conditions by using a machine inspection mode, and performing image enhancement to obtain a data set; performing category and fault marking on to-be-detected parts in each image in the data set, and training a detection model based on deep learning by using the marked images to obtain a trained detection model; and performing fault detection on the power transmission line by using the trained detection model. The method is mainly used for power transmission line fault detection.

Description

Technical Field

[0001] The present invention relates to the field of transmission line inspection. Background Art

[0002] Transmission lines are the hubs of power transmission. Ensuring the operational reliability and safety of transmission lines is crucial for the national livelihood. However, due to the particularity of the power transmission and distribution system, it is usually built in complex terrain areas such as forests and mountains. It is not only affected by the complex and changeable external environment, but also interfered by factors such as its own structural design and external force damage, and various faults are likely to occur. Therefore, regular safety inspections of transmission lines have become an important task for the safe operation of transmission lines.

[0003] Transmission line inspection refers to regularly checking the infrastructure of transmission lines, including line equipment such as iron towers, conductors, insulators, and connecting fittings, to identify and detect (locate) potential problems that may affect the safety and reliability of the power grid. Usually, the inspection of transmission lines is carried out by manual operators on foot or using ground vehicles.

[0004] In the current implementation methods of intelligent inspection of transmission lines, the more common fault monitoring methods include: traditional manual line inspection, unmanned aerial vehicle (UAV) inspection method, infrared thermal imaging method, laser scanning method, sensor monitoring method, cloud computing and big data analysis, vibration monitoring technology, GIS (Geographic Information System), etc. Among them, the transmission line inspection method based on image perception technology is relatively advanced. Its advantages lie in the efficient automation, accurate fault detection, real-time data processing, etc. in the inspection process, so as to reduce the labor cost by improving the degree of intelligence and ensure the safe and stable operation of the power system. Such a system can collect the equipment status information of the transmission line and can resist the interference of complex external factors such as rain, snow, haze, wind, and earthquake. First, image information of the high-voltage transmission line equipment status is collected by inspection equipment such as UAVs, and then the collected information is transmitted to the ground image monitoring center through a wireless communication network. Various status and data sets in different environments are batch-organized and transmitted to the server for data preprocessing and data enhancement operations. Finally, image analysis is performed in cooperation with relevant algorithms to complete the data training and testing of the target detection task.

[0005] In summary, when traditional image processing methods are applied to transmission line inspection tasks, since the detection and segmentation methods are greatly affected by factors such as image background and brightness distribution, and the traditional methods lack a large amount of prior knowledge and cannot cope with complex and changeable scene environments, it is necessary to perform deep feature extraction on complex image environments through deep learning methods to improve the generalization ability of the inspection task for multiple targets. Summary of the Invention

[0006] The object of the present invention is to solve the problem that in the existing traditional image processing methods applied to the inspection task of transmission lines, due to the large influence of factors such as image background and brightness distribution on the detection and segmentation methods, and the lack of a large amount of prior knowledge in traditional methods, it is impossible to cope with complex and changeable scene environments, and a transmission line intelligent inspection method based on image perception is provided.

[0007] A transmission line intelligent inspection method based on image perception, the method includes:

[0008] Collect images of transmission line components under different meteorological conditions by means of machine inspection, and perform image enhancement to obtain a data set;

[0009] After classifying and fault-marking the components to be detected in each image in the data set, use the marked images to train a detection model based on deep learning to obtain a trained detection model;

[0010] Use the trained detection model to detect faults in the transmission line.

[0011] Preferably, the categories of components to be detected include conductors, insulators and connecting fittings.

[0012] Preferably, the conductor fault categories include conductor wind deflection, conductor strand breakage and conductor icing; the insulator fault categories include white explosion, breakage and pollution;

[0013] The connecting fitting faults include pin bolt faults, vibration damping hammer faults and grading ring faults.

[0014] Preferably, the transmission line intelligent inspection method based on image perception further includes the step of expanding the data set, specifically:

[0015] Extract features from the collected images, and generate new samples by means of interpolation, sampling and synthesis; wherein, each image is used as a sample.

[0016] Preferably, the machine for inspection is one or more of an inspection UAV, a line inspection robot and a tower-mounted camera.

[0017] Preferably, the transmission line intelligent inspection method based on image perception further includes the step of preprocessing the images in the data set, specifically:

[0018] Process the images in the data set through image super-resolution and deblurring algorithms.

[0019] Preferably, the detection model based on deep learning is implemented using the YOLOv7 neural network.

[0020] Advantages of the present invention:

[0021] In order to ensure that the training model is more robust, the present invention performs image enhancement on the collected original images, and extracts deep features from complex image environments through deep learning methods to improve the generalization ability of the inspection task for multiple targets, avoiding the defects of traditional image processing methods applied to transmission line inspection tasks. Due to the large influence of factors such as image background and brightness distribution on the detection and segmentation methods, the traditional methods lack a large amount of prior knowledge and cannot cope with complex and changeable scene environments.

[0022] The present invention also expands the data to address the problem of scarce training samples in outdoor complex environments and cope with complex and changeable scene environments; and uses different inspection methods for different target categories to ensure the effectiveness of image acquisition. The present invention is mainly used for transmission line fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart of the intelligent inspection method for transmission lines based on image perception according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0026] See Figure 1 To illustrate this embodiment, the intelligent inspection method for transmission lines based on image perception described in this embodiment includes:

[0027] Collect images of transmission line components under different meteorological conditions by means of machine inspection, and perform image enhancement to obtain a data set;

[0028] After classifying and fault-marking the components to be detected in each image in the data set, use the marked images to train a detection model based on deep learning to obtain a trained detection model;

[0029] Use the trained detection model to detect faults in the transmission line.

[0030] Among them, the categories of components to be detected include conductors, insulators, and connecting fittings.

[0031] Specifically, the wire fault categories include wire wind deflection, wire strand breakage, and wire icing; the insulator fault categories include white explosion, damage, and pollution; the connection fitting faults include pin and bolt faults, damper faults, and grading ring faults.

[0032] In this embodiment, in order to ensure that the training model is more robust, the original images collected are subjected to image enhancement. The specific image enhancement methods are operations such as mirroring, flipping, scaling, and adding noise to enhance the data set. The final data set is composed of the original images and the enhanced images.

[0033] Deep learning learns image features and automatically extracts defect information in transmission line images through multi-layer convolution and pooling operations. Therefore, in this embodiment, a detection model based on deep learning is used to perform deep feature extraction on a complex image environment to improve the generalization ability of the inspection task for multiple targets, avoiding the defects that traditional image processing methods are greatly affected by factors such as image background and brightness distribution in the application of transmission line inspection tasks, and the traditional methods lack a large amount of prior knowledge and cannot cope with complex and changeable scene environments.

[0034] Regarding the problem of lack of training samples in outdoor complex environments, for example, if there are faults in transmission line facilities built on canyon cliffs and steep mountain peaks, it will be extremely difficult to conduct inspection work and obtain original data. At this time, it is necessary to expand the data through certain algorithms based on a small amount of precious original data. Therefore, the intelligent inspection method for transmission lines based on image perception described in the present invention further includes the step of expanding the data set, specifically:

[0035] Extract features from the collected images and generate new samples through interpolation, sampling, and synthesis; where each image is used as a sample.

[0036] Furthermore, during the operation and maintenance of the power system, common component defects of transmission lines mainly include wire faults, insulator faults, connection fitting faults, etc. The inspection methods used for different target categories are not the same due to differences in shape and usage characteristics. Common wire faults are usually monitored by drone line inspection, and in special cases such as wire wind deflection faults, they are completed by line inspection robots; while insulator faults and connection fitting corrosion faults are usually captured by drone hovering under suitable meteorological conditions. However, limited by environmental factors, there are problems such as blurring, low resolution, backlighting, and limited photographing angles in some drone aerial images, which will all affect the detection accuracy of the model. Therefore, a suitable meteorological environment is selected for inspection operations and image acquisition.

[0037] Furthermore, data augmentation techniques can be utilized to expand the training data, enhancing the model's robustness in identifying defects in transmission lines under complex backgrounds. Specifically, images can be processed through image super-resolution and de-blurring algorithms to provide an accurate data basis for improving the detection accuracy of the trained detection model.

[0038] Furthermore, the detection model based on deep learning is implemented using the YOLOv7 neural network.

[0039] Although the present invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed, as long as they do not depart from the spirit and scope of the invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein can be combined in a manner different from that described in the original claims. It should also be understood that the features described in connection with a single embodiment can be used in other described embodiments.

Claims

1. An intelligent inspection method for transmission lines based on image perception, characterized in that The method includes: Collecting images of transmission line components under different meteorological conditions by means of machine inspection, and performing image enhancement to obtain a dataset; After classifying and fault-marking the components to be detected in each image in the dataset, using the marked images to train a deep learning-based detection model to obtain a trained detection model; Using the trained detection model to detect faults in the transmission line.

2. The intelligent inspection method for transmission lines based on image perception according to claim 1, wherein The categories of components to be detected include conductors, insulators, and connecting fittings.

3. The intelligent inspection method for transmission lines based on image perception according to claim 2, wherein The conductor fault categories include conductor wind deflection, conductor strand breakage, and conductor icing; the insulator fault categories include white explosion, breakage, and contamination; The connecting fitting faults include pin bolt faults, damper faults, and grading ring faults.

4. The intelligent inspection method for transmission lines based on image perception according to claim 1, wherein It also includes the step of expanding the dataset, specifically: Performing feature extraction on the collected images, and generating new samples through interpolation, sampling, and synthesis; where each image is used as a sample.

5. The intelligent inspection method for transmission lines based on image perception according to claim 1, wherein The machine for inspection is one or more of an inspection UAV, a line inspection robot, and a tower-mounted camera.

6. The intelligent inspection method for transmission lines based on image perception according to claim 1, characterized in that It also includes the step of preprocessing the images in the dataset, specifically: Processing the images in the dataset through image super-resolution and deblurring algorithms.

7. The intelligent inspection method for transmission lines based on image perception according to claim 1, wherein The deep learning-based detection model is implemented using the YOLOv7 neural network.

Images