Insulation Defects Research Articles

On image transformation for partial discharge source identification in vehicle cable terminals of high‐speed trains

AbstractPartial discharge (PD) detection of cable terminals is crucial for the safe operation of the traction power system in trains. However, similar PD signals in complex train‐operating environments cause difficulty to recognise the insulation defects. Therefore, a PD signal image transformation recognition method is proposed for PD detection of cable terminal defects to identify defects in cable terminals with similar PD characteristics accurately. In the proposed method, the raw PD signals are firstly transformed to images via the Gramian angular field (GAF) representation. This can reveal the discriminative characteristics embedded in the original PD signals and subsequently facilitate differentiating the PD sources, which exhibit similar characteristic in the time domain. The obtained GAF representation of PD signals (named as PD GAF images) is extracted from local and global features to train an efficient MobileVIT model, which is then utilised to identify similar types of PD sources in cable terminals. The results show that the proposed method achieves 97.5% recognition accuracy in the field experiment, which is superior to other methods.

Effect of micrometer bubble interfaces on the detection and measurement method of XLPE

Hidden air bubble defects in the XLPE (Cross-Linked Polyethylene) insulation will seriously affect its electrical performance, which needs to be detected timely to ensure the safe and stable operation of the power system, but the interfaces between bubbles and XLPE may affect the detection. This paper proposes a novel method for high-precision detection and measurement method of micron-sized bubbles in XLPE based on THz time-domain reflection technology. To analyze the influencing factors limiting the high-precision detection and measurement of micron-sized defects, firstly, the propagation law of THz wave in XLPE with and without defects is studied, and the unipolar and bipolar pulses reflected at the defects are analyzed by overlap, and then a new method of micron-sized defects detection and measurement is proposed based on the overlap property of the bipolar pulses, and the relationship between the minimum detecting frequency and the size of the defects in the various detecting methods is determined. And the finite element method is used to simulate the THz detection of XLPE samples with defect thicknesses of 100, 7, 5, and 3 μm, respectively, and the proposed method is validated in conjunction with experiments. The results show that micrometer defects can be detected and measured with high precision using this proposed method, and the error can be controlled within 4.39%, which is significantly better than the conventional detection method. The proposed method can provide a new idea for non-destructive testing of insulation defects.

Improving Energy Storage Density of Biaxially Oriented Polypropylene Film by Quickly Repairing Surface Insulation Defects

Improving Energy Storage Density of Biaxially Oriented Polypropylene Film by Quickly Repairing Surface Insulation Defects

Decision Tree Clusters: Non-destructive detection of overheating defects in porcelain insulators using quantitative thermal imaging techniques

Porcelain insulators are subject to performance deterioration during operation under the influence of complex environments, thus increasing the occurrence of flashover accidents. Regular inspection of insulator condition is of great significance for the stable operation of power grids. Therefore, we propose a non-destructive detection method of insulator overheating defects using quantitative thermography. It can be used for defect detection of porcelain insulators with internal overheating defects. In this study, the relationship between the insulating properties of porcelain insulators and the heating anomalies was first analyzed. Then, Decision Tree Clusters (DTC) algorithm is used to extract the spatial temperature feature information to achieve the defective insulator location capture. Finally, Submodular-pick Local Interpretable Model-agnostic Explanations (SLIME) is used for model decision visualization to verify the feasibility of the technique. The experimental results show that DTC not only ensures the accuracy of insulator long-distance detection, but also realizes the quantitative detection of insulators.

Effect of Interface Defects on the Electric–Thermal–Stress Coupling Field Distribution of Cable Accessory Insulation

The combined insulation interface of a high-voltage cable and accessories is the weakest part of a cable system. In this paper, the parameters of the dielectric constant, thermal conductivity, and elastic modulus of cross-linked polyethylene (XLPE) and silicone rubber (SIR) are obtained experimentally. On this basis, the model of a specific type of 110 kV cable and prefabricated insulation joint is established. A simulation of the electric–thermal–stress coupling field in the presence of typical defects in the main insulation–inner semi-conductive (SEMI) shielding layer (XLPE/SEMI interface) and the main insulation–silicone rubber insulation layer (XLPE/SIR interface) is studied. The simulation results show that at the XLPE/SIR interface, the electric field distortion caused by bubble defects reached 20.17 kV/mm, and the temperature rose to 56.15 °C. The effect of air-gap defects on the interface is similar to that of bubble defects. In addition, the semi-conductive impurity defects induced an increase in temperature to 56.82 °C and an increase in stress to 0.32 MPa. At the XLPE/SEMI interface, the electric field distortion induced by bubble defects was 19.98 kV/mm, and the temperature rose to 61.72 °C. The electric field distortion caused by metallic and semi-conductive defects was 8.44 kV/mm and 8.64 kV/mm, respectively. This study serves as a reference for the fault analysis and the operation and maintenance of cable accessories.

Target localization and defect detection of distribution insulators based on ECA‐SqueezeNet and CVAE‐GAN

AbstractInsulators, as typical equipment for distribution networks, provide good electrical insulation between live conductors and earth. Timely and accurate detection is essential for insulator detection issues. However, as the complexity of neural networks increases, the detection efficiency is often lower. Therefore, this paper proposes a fast insulator positioning and defect detection method. Firstly, for insulator target localization, the SqueezeNet network is improved using ECA attention mechanism. In addition, to address the issue of low defect detection accuracy, a joint algorithm has been proposed. The integration of convolutional variational autoencoder (CVAE) and generative adversarial network (GAN) solve their own shortcomings due to different image focus angles. The target localization accuracy reaches 94.30%, and the defect detection accuracy reaches 89.60%. It solves the problems of difficulty in locating small targets in a large field of view and inaccurate detection due to a small number of abnormal samples. This method has been tried and tested in practical distribution network systems.

PAL-YOLOv8: A Lightweight Algorithm for Insulator Defect Detection

To address the challenges of high model complexity and low accuracy in detecting small targets in insulator defect detection using UAV aerial imagery, we propose a lightweight algorithm, PAL-YOLOv8. Firstly, the baseline model, YOLOv8n, is enhanced by incorporating the PKI Block from PKINet to improve the C2f module, effectively reducing the model complexity and enhancing feature extraction capabilities. Secondly, Adown from YOLOv9 is employed in the backbone and neck for downsampling, which retains more feature information while reducing the feature map size, thus improving the detection accuracy. Additionally, Focaler-SIoU is used as the bounding-box regression loss function to improve model performance by focusing on different regression samples. Finally, pruning is applied to the improved model to further reduce its size. The experimental results show that PAL-YOLOv8 achieves an mAP50 of 95.0%, which represents increases of 5.5% and 2.6% over YOLOv8n and YOLOv9t, respectively. Furthermore, GFLOPs is only 3.9, the model size is just 2.7 MB, and the parameter count is only 1.24 × 106.

Semi-supervised learning for gas insulated switchgear partial discharge pattern recognition in the case of limited labeled data

Semi-supervised learning has better and more efficient performance than supervised algorithms in the case of limited labeled data. Existing methods for diagnosing partial discharge (PD) insulation defects in gas-insulated switchgear (GIS) equipment can only be effective if there is sufficient labeled data. However, in the actual working conditions of GIS equipment, insulation defect data is very scarce, where labeled data is more expensive to obtain and most of the data is unlabeled. In the case of limited PD labeled data, it is still a serious challenge to achieve higher classification accuracy of GIS PD pattern recognition. Therefore, we propose a semi-supervised self-training algorithm based on density peaks of local neighbor Information. Firstly, an improved density peak clustering algorithm based on local neighbor information is proposed, which no longer depends on the truncation distance, and considers the local information to better reflect the local density. Secondly, using local neighbor Information of labeled data, the criterion of confidence of unlabeled data is improved. Then, the PD unlabeled data with pseudo-labels are used to build a strong classifier for GIS PD pattern recognition. The experimental results show that the proposed algorithm has higher classification accuracy than other semi-supervised algorithms. When the proportion of labeled data is 10 %, the recognition accuracy can reach 65.98 %, which is the highest in the comparison algorithm. The proposed algorithm provides a feasible solution for GIS PD pattern recognition in the case of limited labeled data.

Insulator Defect Detection Based on YOLOv5s-KE

To tackle the issue of low detection accuracy in insulator images caused by intricate backgrounds and small defect sizes, as well as the requirement for real-time detection on embedded and mobile devices, this research introduces the YOLOv5s-KE model. Integrating multiple strategies, YOLOv5s-KE aims to boost detection accuracy significantly. Initially, an enhanced anchor generation method utilizing the K-means++ algorithm is proposed to generate more appropriate anchor boxes for insulator defects. Moreover, an attention mechanism is integrated into both the backbone and neck networks to enhance the model’s capacity to focus on defect features and resist interference. To improve the detection of small defects, the EIoU loss function is implemented in place of the original CIoU loss function. In order to meet the real-time detection needs on embedded and mobile devices, the model is further refined through the integration of Ghost convolution for lightweight feature extraction and a linear transformation to reduce the computational burden of standard convolution. A channel pruning strategy is deployed to optimize the sparsely trained network, diminishing redundancy, and improving model generalization. Additionally, the CARAFE operator replaces the original upsampling operator to minimize model parameters and elevate detection speed. Experimental outcomes demonstrate that YOLOv5s-KE achieves a detection accuracy of 92.3% on the Chinese transmission line insulator dataset, marking a 5.2% enhancement over the original YOLOv5s. The streamlined version of YOLOv5s-KE achieves a detection speed of 94.3 frames per second, indicating an improvement of 30.1 frames per second compared to the original model. Model parameters are condensed to 9.6 M, resulting in a detection accuracy of 91.1%. This study underscores the precision and efficiency of the proposed approach, suggesting that the advanced strategies explored introduce novel possibilities for insulator defect detection.

Insulator defect detection algorithm based on YOLOv8-ISDD

Abstract Insulators are widely used in transmission lines due to their good insulation performance and pollution resistance. The damage to insulators will affect their service life and insulation performance, which brings potential safety hazards to transmission lines. With the replacement of manual inspection with UAV inspection, image detection technology based on deep learning is gradually combined with UAV inspection. Usually, the insulator defect target accounts for a small proportion of the acquired image. Therefore, the manuscript introduces an enhanced algorithm for detecting flaws on the surface of insulators, designated as YOLOv8-ISDD. In the backbone network, the YOLOv8-ISDD algorithm introduces reparameterized convolution (Repconv) to enhance the network’s capability in extracting features of defects; to fuse the features of different resolutions, PANet in the neck network is replaced by Bi-FPN for integrating features. Within the head network, the multi-scale detection head output is used to increase the diminutive object identification stratum and boost the algorithm’s capability to discern lesser targets. Experiments show that the mAP@0.5 of the YOLOv8-ISDD algorithm is 88.9%, which is 6% higher than that of the original YOLOv8 algorithm, indicating that YOLOv8-ISDD can detect insulator defects more accurately.

A Lightweight Insulator Defect Detection Model Based on Drone Images

With the continuous development and construction of new power systems, using drones to inspect the condition of transmission line insulators has become an inevitable trend. To facilitate the deployment of drone hardware equipment, this paper proposes IDD-YOLO (Insulator Defect Detection-YOLO), a lightweight insulator defect detection model. Initially, the backbone network of IDD-YOLO employs GhostNet for feature extraction. However, due to the limited feature extraction capability of GhostNet, we designed a lightweight attention mechanism called LCSA (Lightweight Channel-Spatial Attention), which is combined with GhostNet to capture features more comprehensively. Secondly, the neck network of IDD-YOLO utilizes PANet for feature transformation and introduces GSConv and C3Ghost convolution modules to reduce redundant parameters and lighten the network. The head network employs the YOLO detection head, incorporating the EIOU loss function and Mish activation function to optimize the speed and accuracy of insulator defect detection. Finally, the model is optimized using TensorRT and deployed on the NVIDIA Jetson TX2 NX mobile platform to test the actual inference speed of the model. The experimental results demonstrate that the model exhibits outstanding performance on both the proprietary ID-2024 insulator defect dataset and the public SFID insulator dataset. After optimization with TensorRT, the actual inference speed of the IDD-YOLO model reached 20.83 frames per second (FPS), meeting the demands for accurate and real-time inspection of insulator defects by drones.

Adsorption and sensing properties of TiO2-modified MoSe2 monolayer for SF6 decomposition components based on the first-principles

As crucial components in power systems, SF6 gas-insulated equipment inevitably develops insulation defects and faults over long-term operation, leading to the decomposition of SF6 gas. Therefore, detecting SF6 decomposition components is essential for assessing the operational status of SF6 gas-insulated equipment and ensuring the safe and stable operation of power systems. Based on first principles, the modification mechanism of TiO2 on MoSe2 was explored, and optimal adsorption models for gases SO2, H2S, SOF2, and SO2F2 on this composite material were established. The adsorption characteristics of each system were analyzed through parameters such as adsorption energy, adsorption distance, charge transfer, electron density difference, and density of states, and the sensing properties of the systems were discussed using parameters including changes in frontier molecular orbitals, desorption time, and sensitivity. The results indicate that TiO2–MoSe2 has difficulty capturing SO2F2, but exhibits chemisorption for SO2, H2S, and SOF2 with adsorption energies of −1.25 eV, −0.99 eV, and −3.40 eV, respectively. Upon adsorption of SO2, TiO2–MoSe2 shows a significant change in electrical conductivity, a desorption time of 4.51 s (at 498 K), and a strong sensitivity response. However, the response to H2S is relatively weak, with lower sensitivity. Therefore, this material is suitable for detecting SO2 but not H2S. TiO2–MoSe2 demonstrates a significant response and excellent sensitivity to SOF2; however, the adsorption process lacks repeatability. Therefore, it can be used as a disposable gas sensor or solid adsorbent for SOF2 detection. The results of this study provide theoretical support for the application of MoSe2-based sensors in the monitoring and fault diagnosis of SF6 gas-insulated equipment.

Online Monitoring Technology for Cable Insulation in Distribution Networks Based on Line Conduction Characteristics

Due to the increasing load of the power grid and the rapid expansion of cable line construction, monitoring cable insulation performance has become very important. To address this challenge, this paper proposes an online monitoring system for cable insulation in distribution networks based on line conduction characteristics. The system combines advanced signal acquisition technology, data processing methods, and deep learning (DL) algorithms to enable real-time monitoring and accurate evaluation of cable insulation status. The results demonstrate that the system excels in real-time monitoring of cable insulation status. Compared to traditional methods, the system proposed in this paper shows significant advantages in key indicators such as missed detection rate, fault location accuracy, recognition speed, and recall rate. Specifically, the system effectively reduces the missed detection rate, improves the accuracy of fault location, accelerates identification speed, and enhances the recall rate, enabling more comprehensive detection of insulation faults.

Research on insulator image segmentation and defect recognition technology based on U‐Net and YOLOv7

AbstractThis study focuses on aerial images in power line inspection, using a small sample size and concentrating on accurately segmenting insulators in images and identifying potential “self‐explode” defects through deep learning methods. The research process consists of four key steps: image segmentation of insulators, identification of small connected regions, data augmentation of original samples, and detection of insulator defects using the YOLO v7 model. In this paper, due to the small sample size, sample expansion is considered first. A sliding window approach is adopted to crop images, increasing the number of training samples. Subsequently, the U‐Net neural network model for semantic segmentation is used to train insulator images, thereby generating preliminary mask images of insulators. Then, through connected region area filtering techniques, smaller connected regions are removed to eliminate small speckles in the predicted mask images, obtaining more accurate insulator mask images. The evaluation metric for image recognition, the dice coefficient, is 93.67%. To target the identification of insulator defects, 35 images with insulator defects from the original samples are augmented. These images are input into the YOLO v7 network for further training, ultimately achieving effective detection of insulator “self‐explode” defects.

Electrical and electrochemical properties of A-site non-stoichiometry Ca0.67La0.22□0.11Ti(1−x)CrxO3−δ electrolyte materials for solid oxide fuel cells

Due to increasing global energy demands and environmental concerns, the search for alternative and environmentally friendly energy sources is of paramount importance. Solid oxide fuel cells (SOFCs) have emerged as a promising solution thanks to their high efficiency and minimal environmental impact. In this regard, perovskite oxides, with their unique properties, have attracted significant attention. This study investigates the dielectric dispersion, electrical features, scaling behavior, and optical defects of Ca0.67La0.22Ti0.85Cr0.15O3 with x = 0.15, CLT0.85Cr0.15. The presence of a vacancy in the A-site is denoted by the square in the formula. Relaxation phenomena were analysed using dielectric and modulus formalism, while the conductivity mechanism was explored through electrical conductivity measurements. The optical defects of CLT0.85Cr0.15 were analysed using electron paramagnetic resonance (EPR) spectroscopy. The findings revealed the formation of Cr3+–VO center defects. These defects serve as sources of in-gap electron traps, thereby enhancing the optical properties of CLT0.85Cr0.15, making it as a compelling material for various optical applications. The power density increased with temperature, reaching 0.73 W/cm2 at 850 °C, suggesting that CLT0.85Cr0.15 could be a promising electrolyte for IT-SOFCs. The peak in power density correlated with temperature due to thermal effects on ion motion. However, the open circuit voltage decreased with temperature increase, due to increased oxygen vacancies and electron/hole conduction in CLT0.85Cr0.15. The findings suggest that perovskite materials could revolutionize SOFCs technology, paving the way for more efficient and environmentally friendly energy solutions.

Locating Insulation Defects in HV Substations Using HFCT Sensors and AI Diagnostic Tools.

In general, a high voltage (HV) substation can be made up of multiple insulation subsystems: an air insulation subsystem (AIS), gas insulation subsystem (GIS), liquid insulation subsystem (power transformers), and solid insulation subsystem (power cables), all of them with their grounding structures interconnected and linked to the substation earth. Partial discharge (PD) pulses, which are generated in a HV apparatus belonging to a subsystem, travel through the grounding structures of the others. PD analyzers using high-frequency current transformer (HFCT) sensors, which are installed at the connections between the grounding structures, are sensitive to these traveling pulses. In a substation made up of an AIS, several non-critical PD sources can be detected, such as possible corona, air surface, or floating discharges. To perform the correct diagnosis, non-critical PD sources must be separated from critical PD sources related to insulation defects, such as a cavity in a solid dielectric material, mobile particles in SF6, or surface discharges in oil. Powerful diagnostic tools using PD clustering and phase-resolved PD (PRPD) pattern recognition have been developed to check the insulation condition of HV substations. However, a common issue is how to determine the subsystem in which a critical PD source is located when there are several PD sources, and a critical one is near the boundary between two HV subsystems, e.g., a cavity defect located between a cable end and a GIS. The traveling direction of the detected PD is valuable information to determine the subsystem in which the insulation defect is located. However, incorrect diagnostics are usually due to the constraints of PD measuring systems and inadequate PD diagnostic procedures. This paper presents a diagnostic procedure using an appropriate PD analyzer with multiple HFCT sensors to carry out efficient insulation condition diagnoses. This PD procedure has been developed on the basis of laboratory tests, transient signal modeling, and validation tests. The validation tests were carried out in a special test bench developed for the characterization of PD analyzers. To demonstrate the effectiveness of the procedure, a real case is also presented, where satisfactory results are shown.

Insulator detection based on FA‐YOLO network with improved feature extraction ability

AbstractUnmanned aerial vehicle insulator detection that aims to recognize defective insulators from transmission lines has made significant progress in recent years. However, it still faces challenges, such as the complex background of aerial images and the small memory of unmanned aerial vehicles. This paper proposes a refined insulator detection algorithm that integrates the attention mechanism in YOLOv8 to improve the feature extraction ability. Specifically, this paper introduces a fast vision transformers structure in the you only look once (YOLO) v8 backbone section to enhance feature extraction by capturing local and global features. Additionally, the global attention mechanism is incorporated in the neck for additional feature extraction by merging comprehensive spatial and channel information into the output. Furthermore, we amalgamate depth‐wise convolution, graph convolution, and residual operation in the global attention mechanism module. This design can mitigate the issues of gradient vanishing or exploding and meanwhile enhance the distinction between spatial attention and channel attention. The proposed model is then applied to a public dataset and a set of real images from a specific power station, and the detection results show that it outperforms many competitors in terms of accuracy, efficiency, and memory size.

Enhanced recognition of insulator defects on power transmission lines via proposal-based detection model with integrated improvement methods

Deep learning-driven transmission line inspection is a critical area for smart power grid development. Despite advances in deep learning for insulator defect detection, challenges remain in model robustness and adaptability for the varying real-world adaptability, especially for insignificant defects in complex backgrounds. This study presents a comprehensive improvement strategy for detecting insulators and cross-scale broken defects on transmission lines, employing a proposal-based detection model. The model introduces a holistic pipeline of improved methods, including backbone modification, anchor box scale recalibration, and improvements in Region of Interest (RoI) downsampling alignment and Intersection over Union (IoU) loss function. Various backbone networks, including convolutional network (ConvNet) and Vision Transformer (ViT) structures, are constructed and integrated with attention modules, specifically designed to amplify the perception of insulators and defective regions. The geometric scale of anchor boxes is reconstructed using a developed clustering algorithm, considering the elongated characteristics of insulator strings to improve the adaptability of anchor boxes. Bilinear interpolation is utilized to mitigate spatial misalignment issues during the downsampling process of Region Proposal Network (RPN)-based proposals. The experimental results indicate that the improved models with the Swin Transformer (Swin-T) backbone framework achieve the mean Average Precision (mAP)@0.5 of 88.42% and mAP@0.7 of 60.52%, with a defect recall rate of 81.94%. Additionally, the improved IoU loss function contributes to the performance of the model at higher IoU thresholds. The results of this study contribute to the further development of defect detection frameworks for power vision applications.

Detection of defects in composite insulators based on laser‐induced plasma combined with machine learning

AbstractDue to Prolonged operation and external environmental factors, composite insulators may develop various defects, which can potentially lead to serious accidents in power systems. Detecting and analyzing these defects is critically important. In this study, we combine machine learning with laser‐induced breakdown spectroscopy (LIBS) to identify defects in composite insulators and obtain spectral data from both defective and nondefective samples. The Uniform Manifold Approximation and Projection algorithm was employed to reduce the dimensionality of the data, thereby enhancing detection accuracy and efficiency. Decision tree, random forest (RF), K‐nearest neighbor, and support vector machine algorithms were used to classify the dimensionality‐reduced data. The results indicate that the RF algorithm achieved the best classification performance, with accuracies of 100% and 95.46% for the training and test sets, respectively. Furthermore, the precision, recall, and F1 scores, which reflect model performance, also showed superior results with the RF algorithm. These findings suggest that the combination of LIBS technology and machine learning can rapidly and accurately detect and analyze defects in composite insulators, offering new insights and methods for defect detection in composite insulators.

Railroad Catenary Insulator Fault Detection Based on Improved Faster R-CNN

Background:: The railroad catenary insulator, which is a crucial component of the catenary system and is situated between the pillar and wrist arm, is crucial for electrical conductor isolation, electrical equipment insulation, mechanical load bearing, anti-fouling, and anti-leakage. The catenary insulators will experience tarnished flash, breakage, insulation strength deterioration, and other issues as a result of the long-term outside unfavorable working circumstances. The train electrical system's ability to operate normally is greatly hampered by these problems. Although there are many patents and articles related to insulator fault detection, the precision is not high enough. Therefore, it is crucial to improve the precision of catenary insulator fault detection. Objective:: An improved region-based convolutional neural networks (Faster R-CNN)-based fault detection method for railway catenary insulators is proposed in response to the long detection time of the conventional railroad catenary insulator fault, the low precision of the catenary insulator fault detection for occlusion and truncation, the poor performance of multi-scale object detection, and the processing of class unbalance problem. Methods:: The Faster R-CNN is optimized from four perspectives: feature extraction, feature fusion, candidate box screening, and loss function, in accordance with the properties of the catenary insulator. First, to solve the problem of multi-scale catenary insulator fault detection, convolutional block attention module (CBAM) and feature pyramid network (FPN) are used to fuse the deep feature and shallow features of the image. This results in a feature map with more critical semantic information and higher resolution. After that, the weighted non-maximum suppression (WNMS) algorithm improved by distance-intersection over union (DIOU) and Gaussian weighting function is used instead of the traditional NMS algorithm, which effectively introduces the overlap of detection frames into the confidence level and makes full use of the effective information of the detection frames. Finally, the improved Focal loss is used as the classification loss, and the focusing parameter and the balance factor of the Focal Loss are adjusted dynamically to solve the problem of sample imbalance and difficult sample identification in the model better. Results:: The effects of SSD, YOLOV3, traditional Faster R-CNN and improved Faster R-CNN models are tested on the contact network insulator fault detection dataset constructed in this paper, and the experimental results show that the improved Faster R-CNN has higher precision, recall, and mAP compared to the other detection models, which reach 94.31%, 96.68% and 95.22%, respectively. Conclusion:: The results of the experiments demonstrate that this method may successfully detect the faults in different scale catenary insulators. It can effectively detect truncated, obscured faulty catenary insulators. It has higher precision and recall and provides a reliable reference for maintaining faulty insulators in railway catenary.