Inspection Speed Research Articles

SELECTION OF A METHOD FOR MONITORING THE PIPELINE FOR THE PRESENCE OF CORROSION DAMAGES

Pitting corrosion is a form of localized destruction of metals and alloys. Leads to the formation of depressions in certain areas of the surface of metal products.This type of metal destruction is unpredictable and the most dangerous:— pitting grows deeper into the metal at a rate exceeding the rate of general corrosion;— rapid development can lead to the formation of corrosion cracks, fistulas or pinpoint through destruction of the metal;— due to its rapid spread and small size, the defect is not easy to detect even with the use of non-destructive testing methods;— with a large accumulation of pitting corrosion, the depth of damage on the metal surface is greater;— pitting type corrosion can spread on stainless steels and in layers of the top anti-corrosion coating.Work has been carried out to select a method for technical diagnostics of the pipeline, capable of identifying the presence of pitting corrosion damage to the pipe.As a sample, a cut-out pipeline element with a diameter of 325 mm, a wall thickness of 5 mm, steel grade TP304 according to ASTM A312 (equivalent to 08Х18Н10) was provided, containing areas affected by internal pitting corrosion of the base metal, through lesions with chipping of the factory welded joint and multiple corrosion lesions lower reinforcement of assembly welded joints.In order to solve the problem, it was recommended to test certain equipment using eddy current testing methods and ultrasonic testing methods.During the work, both welded joints and the base metal of the pipe sample were scanned.The low-frequency eddy current flaw detector TiS 8C allows you to detect localized defects with high productivity. Eddify's Reddy Eddy Current Flaw Detector allows you to detect high throughput localized defects and surface defects in welds.Inspection using ultrasonic phased array flaw detector OmniScan SX is capable of providing high sensitivity, pitting is not detected, inspection speed is low, requires high quality surface preparation and the use of coupling fluidAs a confirmatory testing method, it is recommended to use an ultrasonic thickness gauge or a general-purpose ultrasonic flaw detector.

A self-adaptive co-evolutionary algorithm for multi-objective flexible job-shop rescheduling problem with multi-phase processing speed selection, condition-based preventive maintenance and dynamic repairman assignment

Production scheduling and maintenance planning are two interactive factors in modern manufacturing system. However, at present, almost all studies ignore the impact of unpunctual maintenance activities on the integrated production and maintenance scheduling since the unavailabilities of repairmen are dynamically changed, e.g., repairmen increase, decrease and their unavailable intervals update. Under these contexts, this paper addresses a novel integrated optimization problem of condition-based preventive maintenance (CBPM) and production rescheduling with multi-phase processing speed selection and dynamic repairman assignment. More precisely, (1) a novel multi-phase-multi-threshold CBPM policy with remaining-useful-life-based inspection and multi-phase processing speed selection is proposed to obtain some selectable maintenance plans for each production machine; (2) a hybrid rescheduling strategy (HRS) that includes three rescheduling strategies is designed for responding to the dynamic changes of repairman; and (3) an adaptive clustering- and Meta-Lamarckian learning-based bi-population co-evolutionary algorithm (ACML-BCEA) is developed to deal with the concerned problem. In the numerical simulations, the effectiveness of designed operators and proposed ACML-BCEA algorithm is first verified. Next, the superiority and competitiveness of the proposed CBPM policy and HRS are separately demonstrated by comparing with other CBPM policies and rescheduling strategies. After that, a comprehensive sensitivity analysis is performed to analyze the effect of optional range of processing speed, skill level of selectable repairmen and total number of processing phases, and the analyzing results show that these factors all have a significant impact on the integrated optimization.

Non-destructive evaluation and Structural Health Monitoring of impact induced damage in composite structures using ultrasonic approaches and shearography

Composite structures are massively used thanks to the high strength-to-weight ratio, making them ideal for lightweight and durable structures in aviation. However, Impact may arise at any time during lifetime on composite aerostructures and ensuring their structural integrity via fast inspection is essential to meet airworthiness criteria. Multimodal NDT (Non-Destructive Testing) inspection and characterization of composite structures is a critical process to verify the quality of the composite material and identify any defects. In this context, the paper shows a comparison of several techniques as nondestructive methods: shearography, laser ultrasonics and phased array ultrasonics. In particular, using shearography and laser ultrasonics as non-contact approaches can increase the inspection speed and reliability compared to the state of the art phased array ultrasonics, which is the standard approach used in the aerospace field. For this reason, a sample of interest to the aerospace industry is subject to different low velocity impacts which are characterized by different energies leading to barely visible damage. The results show the capabilities of both approaches in identifying small emerging defects along with the advantage and drawbacks of both methodologies through a critical review of performance thereof evaluating the parameters of their use. In addition, the same approaches are adopted on a stiffener composite plates and compared with continuous health monitoring approaches. In this regard, NDT approaches are demonstrated useful to enhance the SHM warnings.

Development of a comprehensive car safety gears testing procedure for low- to medium-speed lifts in Hong Kong

Operation of the safety gears is the last means of protecting passengers in a high-speed or even freely falling lift car. EN81-20 stipulates that the average retardation when the safety gears are engaged should lie between 0.2 g n and 1.0 g n while g n is the gravitational acceleration. Decades ago, in Hong Kong, to conduct a formal test on the safety gears, the car was fully loaded and the machine brake was manually released to allow the car fall freely until the overspeed governor triggered the engagement of the safety gears. This test was rather vigorous to both the gear jaws and the guiderails. Recently, tests have been conducted with 125% in-car load while the car travels under an inspection (or levelling) speed and the governor is manually triggered. Such a revised test, though much less vigorous, may not be able to verify the average retardation as required. Even a test with full load and free-falling may not reveal the whole truth by merely measuring the resulting contact scratches on the guiderails because a majority of the retardation could have been assisted by the counterweight via the suspension ropes. A method that involves an empty car traveling downward at 1 m/s with the governor manually triggered was first developed in Germany. This method has been improved to form a comprehensive procedure as detailed in this article so that a no-load rated speed, or slightly reduced speed, test is conducted, which is almost non-invasive or much less vigorous. By calculation, the average retardation when all suspension ropes are broken can be estimated. Cross-checking in our experiments revealed the validity of this procedure. The procedure also involves a test to verify the validity of parameters used to perform the estimation and takes care of some modern overspeed governors which cannot be manually triggered. It is hoped that this procedure could encourage maintenance professionals carry out safety gears tests more often to ensure a high safety standard. This comprehensive procedure could encourage more frequent tests on the car safety gears, which is almost non-invasive for manually triggerable overspeed governors or less vigorous for non-triggerable ones. An empty car traveling downward at rated speed, or at slightly reduced speed, is tested by manually triggering the governor while a partially loaded free-falling car is tested if the governor is not manually triggerable. Results can be used to estimate the average retardation of the lift car when all suspension ropes are broken. A pre-test is included to validate critical parameters necessary for the estimation. Experiments conducted so far were on the safety gears of low- to medium-speed lifts. For high-speed lifts beyond 4 m/s, that may be considered in the next stage of research. Although the title of this article includes the city where the procedure was developed, there is no reason why this testing procedure could not be extensively applicable to any traction lift systems around the world at such speed levels.

Evaluation of typical rail defects by induction thermography: experimental results and procedure for data analysis during high-speed laboratory testing

ABSTRACT Rail inspection via non-destructive testing (NDT) techniques is a critical area of research in the railway industry, significantly affecting transport safety and security. Conventional NDT methods face limitations in on-site applications, with emerging techniques improving defect detectability and inspection speed. Recent advances have highlighted infrared thermography, particularly induction thermography, as a promising alternative due to its non-contact, full-field capabilities for detecting both surface and subsurface rail defects. This study explores induction thermography in detecting key defects such as transversal cracks and head checks in different rail tracks. Novel approaches and procedure for data reconstruction that enhance the thermographic inspection results and allow for dynamic testing conditions are proposed. Additionally, the potential for high-speed on-site applications was investigated, utilizing infrared mirrors and optimally shaped coils. Various test parameters, including geometrical resolution, excitation power, and inspection speed up to 20 km/h, were systematically examined.

Multi-Pattern GPU Accelerated Collision-Less Rabin-Karp for NIDS

In the domain of network communication, network intrusion detection systems (NIDS) play a crucial role in maintaining security by identifying potential threats. NIDS relies on packet inspection, often using rule-based databases to scan for malicious patterns. However, the expanding scale of internet connections hampers the rate of packet inspection. To address this, some systems employ GPU accelerated pattern matching algorithms. Yet, this approach is susceptible to denial of service (DOS) attacks, inducing hashing collisions and slowing inspection. This research introduces a GPU-optimized variation of the Rabin-Karp algorithm, achieving scalability on GPUs while resisting DOS attacks. Our open-source solution (https://github.com/AnasAbbas1/NIDS) combines six polynomial hashing functions, eliminating the need for false-positive validation. This leads to a substantial improvement in inspection speed and accuracy. The proposed system ensures minimal packet misclassification rates, solidifying its role as a robust tool for real-time network security.

Pavement Inspection in Transport Infrastructures Using Unmanned Aerial Vehicles (UAVs)

The growing demand for the transportation of goods and people has led to an increasing reliance on transportation infrastructure, which, in turn, subjects the pavements to high traffic volumes. In order to maintain adequate service and safety standards for users, it is essential to establish effective maintenance strategies that ensure the preservation of pavement conditions. As a result, emerging innovations in pavement surface inspection methods, surpassing traditional techniques in terms of inspection and data processing speed and accuracy, have garnered significant attention. One such groundbreaking innovation in inspection systems that has been tested and used in recent years to assess infrastructure condition is the use of unmanned aerial vehicles (UAVs). This study aims to present a critical open-access literature review on the use of UAVs in the inspection of transportation infrastructure pavement in order to assess the type of equipment used, the technology involved, applicability conditions, data processing, and future evolution. The analysis of relevant literature suggests that the integration of intelligent technologies substantially enhances the accuracy of data collection and the detection of pavement distress. Furthermore, it is evident that most applications and research efforts are oriented towards exploring image processing techniques for the creation of 3D pavement models and distress detection and classification.

MRMPro: a web-based tool to improve the speed of manual calibration for multiple reaction monitoring data analysis by mass spectrometry.

As a gold-standard quantitative technique based on mass spectrometry, multiple reaction monitoring (MRM) has been widely used in proteomics and metabolomics. In the analysis of MRM data, as no peak picking algorithm can achieve perfect accuracy, manual inspection is necessary to correct the errors. In large cohort analysis scenarios, the time required for manual inspection is often considerable. Apart from the commercial software that comes with mass spectrometers, the open-source and free software Skyline is the most popular software for quantitative omics. However, this software is not optimized for manual inspection of hundreds of samples, the interactive experience also needs to be improved. Here we introduce MRMPro, a web-based MRM data analysis platform for efficient manual inspection. MRMPro supports data analysis of MRM and schedule MRM data acquired by mass spectrometers of mainstream vendors. With the goal of improving the speed of manual inspection, we implemented a collaborative review system based on cloud architecture, allowing multiple users to review through browsers. To reduce bandwidth usage and improve data retrieval speed, we proposed a MRM data compression algorithm, which reduced data volume by more than 60% and 80% respectively compared to vendor and mzML format. To improve the efficiency of manual inspection, we proposed a retention time drift estimation algorithm based on similarity of chromatograms. The estimated retention time drifts were then used for peak alignment and automatic EIC grouping. Compared with Skyline, MRMPro has higher quantification accuracy and better manual inspection support. In this study, we proposed MRMPro to improve the usability of manual calibration for MRM data analysis. MRMPro is free for non-commercial use. Researchers can access MRMPro through http://mrmpro.csibio.com/ . All major mass spectrometry formats (wiff, raw, mzML, etc.) can be analyzed on the platform. The final identification results can be exported to a common.xlsx format for subsequent analysis.

A stable diffusion enhanced YOLOV5 model for metal stamped part defect detection based on improved network structure

Machine-vision-based defect detection for large metal stamping is a fundamental requirement for improving product quality and inspection speed. However, the performance of machine vision in detecting defects is limited by the large variety of stamped products, significant dimensional differences, and long data-acquisition cycles. To overcome these problems, metal stamping detection has been achieved using the improved YOLOV5 model, which features a slim neck and a multiheaded self-attentive mechanism for detecting wrinkles, holes, and cracks in metal stamping. An image-processing module was used to reduce the impact of metal reflections and improve image quality. Stable diffusion improvement was used to augment the dataset to overcome the small dataset size problem and enhance its generalisation capability. In addition, a BotNet structure with a self-attentive mechanism was introduced into the YOLOV5 model backbone to improve the image feature extraction capability. We then optimised the prediction head structure of YOLOV5 to improve the detection speed and accuracy. Ablation experiments were performed to analyse and verify the effectiveness of each module. The results of the ablation experiments show that the mAP of our proposed stable diffusion improvement data enhancement method, and the YOLO-Bot-VOV algorithm, for metal stamped part defect detection reached 98.2 %, and the parameters were reduced by 0.432 million.

In-Motion Railcar Wheel Inspection using Magnetostrictive EMATs

Due to cyclic mechanical loading, railcar wheels are subject to the development of internal fatigue cracks in the rim underneath the tread surface. Off-line inspections are extremely laborious and require extended out-of-service time, and most existing in-line inspection systems that use conventional ultrasonics or electromagnetic acoustic transducer (EMAT) techniques have several drawbacks that limit their full-scale deployment in service. This paper discusses the work performed on the initial research and development of a proof-of-concept novel magnetostrictive EMAT sensor for in-motion railcar wheel inspection. Wheelsets with known internal discontinuities and a 3.7 m long panelized track were used to demonstrate the feasibility of the developed approach and methods. The results obtained from these tests have shown that the magnetostrictive EMAT sensor that generates 5 mm wavelength shear horizontal waves detects both surface defects and internal fatigue cracks with an excellent signal-to-noise ratio. Different wavelengths can also be added to the sensors to enhance or complement the detection of fatigue cracks at different depths. A total of four or five sensors located at standard concrete crosstie spacing along each rail would be able to provide complete coverage on wheels ranging from 711 to 965 mm in diameter with ±90° wheel coverage per sensor, and a theoretical inspection speed of 56 kph can be achieved using this approach.

Research on the technology of casing inspection through tubing in the Saertu Development Zone of Daqing Oilfield

With the further development of Daqing Oilfield, the occurrence of widespread casing damage in the Saertu Development Area has been on the rise annually.In order to promptly prevent, detect and address casing damage, the management departments of the oil extraction plants must schedule regular casing inspection operations. This approach not only raises production costs but also prolongs the well shutdown period, further impacting crude oil output.This article focuses on the technical needs of oilfield development and has developed tubing and casing detection technology, the radial magnetic scanning system is integrated with the sixteen arm wellbore diameter, the synchronous tubing and casing inspection instrument represents the latest in concurrent inspection technology, developed specifically for the technical demands of oilfield development Without displacing the tubing column, it simultaneously captures data on tubing inner diameter, tubing continuity, the extent of casing damage via magnetic scanning, and continuous 3D imaging of the tubing at standard inspection speeds, the workload of checking and matching tasks were significantly reduced. Between 2022 and 2023, the through-tubing casing inspection technology was utilized in 525 wells, with 356 wells verified by mobilizing the tubing column, achieving a predictive accuracy of 87.6% and generating economic benefits of 69.6 million yuan.

IDP-Net: Industrial defect perception network based on cross-layer semantic information guidance and context concentration enhancement

Applications in Engineering: In industry, surface defect detection is crucial for improving product quality. However, there are many challenges in industrial inspection scenarios, such as interference from background noise, complex small-target problems, significant variations in target objects, and the problem of finding a balance between inspection speed and accuracy. To address the above problems, this paper proposes an industrial defect-aware network based on cross-layer semantic information guidance and contextual attention enhancement (IDP-Net). Specifically, IDP-Net has four different new features. The contribution of artificial intelligence: Firstly, to solve the industrial surface context and defect similarity problem, this paper proposes a Lightweight Local Global Feature Extraction Network (LLG-Net), unlike other methods, the effective combination of self-attention blocks and convolution blocks ensures gradual integration of global and local features across multiple layers, to improve the detection ability of targets with significant changes in scale, this paper designs a Multiscale Perceptual Feature Aggregation Network (MPA-Net), adequately fuses the shallow fine-grained information and the deep semantic information. Then, to enhance the connection between multi-scale semantic information, an adaptive cross-layer feature fusion module (ACFF) is proposed, which is novel in integrating the characteristics of multiple adjacent levels to help the model better capture the different scale characterisation of the target. Finally, a Region Attention Module (RAM) is proposed and introduced in the detector to enhance the attention to the critical regions around the target object. In particular, this paper proposes a new localisation loss function (MEIoU) that enhances the network’s attention to objects at different scales. The experimental results show that 94.3%, 98.7% and 99.5% of mAP@.5 are obtained on steel, PCB and aluminium surface defect datasets, respectively, and 50 FPS is achieved, which is better than the current mainstream detectors and meets the demand of practical industrial production.

Residual shape adaptive dense-nested Unet: Redesign the long lateral skip connections for metal surface tiny defect inspection

The state-of-the-art metal surface defect inspection methods have two problems: (1) they are sensitive to tiny defects because of their extreme small sizes, and (2) they cannot accurately locate the random appeared defects whose semantic relationship with the background context is weak. To solve these problems, Residual Shape Adaptive Dense-nested Unet, a pixel-based defect inspection method is proposed, to obtain the exact shape and location of the defect, by (1) assembling different depth Unet branches with dense skip connections as the feature extractor to combine multi-semantic level visual features; (2) adding Residual Shape Adaptive modules on the dense skip connections to help the model locate the defect regions; and (3) introducing the multi-branch training method which enables model pruning to reduce redundant parameters and accelerate the inspection speed. Experiments are conducted and demonstrated that the Residual Shape Adaptive Dense-nested Unet achieves the best performance among the state-of-the-art defect inspection methods.

Ultrasonic Spot Weld inspection system based on Industrial Robotic, Artificial Intelligence and Artificial Vision

Nowadays, spot welding is still one of the most frequently used techniques for bonding metal plates, for example, in an automotive chassis between 3000 and 4000 spot welds can be found. Using this technique is possible to obtain at a low cost, easily automated and versatile welds with high quality and speed of operation but, the inspection and quality control process needs to be improved so that it is possible to reach automatic systems that allow maximizing the speed of inspection and guarantee, in turn, the repeatability and accuracy of the results, eliminating possible deviations introduced by the operator. In this paper an inspection system, based on the ultrasonic propagation, is described. The main challenges faced are presented, such as the location of the spot weld, using artificial vision, the definition and positioning of the probes until a reliable signal, and the final analysis through a development based on artificial intelligence. The results achieved in real parts, have shown that this robotic system, has a better defect detection capacity than the systems currently on the market, and is applicable to the inspection on assembly lines, optimizing not only the inspection times but also improving the results and therefore the quality of the process.

Design of a laboratory scale Automatic Optical Inspection prototype system for scanning of conveyor belt surfaces — A case study

The design and assembly of an Automatic Optical Inspection (AOI) system is a complex process for any production scenario that considers process quality outcome and loss as an important aspect. This paper discusses the design and implementation of a laboratory-based AOI system through inspection parameter optimization for monitoring the surface condition of conveyor belt segments. The system utilizes a low-cost web camera and a line laser projection device as sensor components in a non-intrusive setup to capture color and depth images of the belt surface. A mixed-level full factorial design of experiment (DoE) approach is devised to examine how different control factors, which in this case are the inspection parameters, such as- the surface illumination, inspection speed, imaging distance, camera frame rate, and camera field of view, affect the response factor, which is the measurement error percentage of the AOI system. The error percentage is obtained by comparing the inspection results of a belt region of known dimensions with corresponding manual measurements. Comprehensive data collection runs followed by the factorial analysis revealed that all individual inspection parameters were significant except for the interaction between imaging distance and field of view. Based on the range of inspection parameter settings investigated, the optimal settings were found to be an imaging distance of 60 cm, a field of view of 75, illumination of 0 lux, an inspection speed of 9 cm/s, and a camera frame rate of 75 frames per second to achieve an inspection error of 0.26%, which is acceptable for our investigation.

Совершенствование грузовой и коммерческой работы на железнодорожном транспорте: концепция и инструментарий

Purpose: The article offers a solution to the problem of low efficiency of freight and commercial work at railway stations with a small loading volume. The goals and objectives of the national project “Digital Economy” and the program “Digital Railway” are considered. The methods of improving freight and commercial work currently used in railway transport to increase the productivity of stations and to increase workload without capital investments in infrastructure are outlined. A prospective algorithm for processing wagons on departure has been formed. The concept of improving the technology of receiving a wagon for transportation is proposed. An intelligent system for receiving wagons for transportation using a neural network has been developed. The results of the development of a program for determining the types of rolling stock in the image are shown. Methods: Methods of analysis, synthesis and neural networks, the theory of automatic control are used. Results: A solution to the problem of increasing the efficiency of the technology of receiving wagons for transportation at stations with insignificant loading volumes is proposed. Practical significance: The introduction of the developed intelligent system at railway stations will improve the quality and speed of commercial inspection when accepting a car for transportation.

Using a magnetic flux leakage method to estimate railway track bolted joint gaps

The periodic estimation of the rail joint gap values allows evaluating the reliability of the continuous welded railway track and preventing its temperature deformations. The purpose of this study is to search for a simple and reliable way to evaluate the rail joint gap, which allows you to determine the desired values during high-speed inspection in automatic mode in a wide range of temperatures. A brief overview of existing techniques for evaluating rail joint gaps based on different physical principles is given. In the proposed technique, a magnetic flux leakage (MFL) method with the placement of electromagnets on the wheel axes of a car bogie is used. Such magnetization systems are applied recently on flaw-detector cars and provide a stable magnetic flux in the tested rails. Using a three-dimensional model of the rail segment with a bolt joint, we studied the characteristics of the magnetic flux leakage fields when changing the size of the rail gap in the entire practical range. The characteristics are selected that are most sensitive to a change in the gap and by which its value can be determined. With small values of the gaps the amplitude of the magnetic sensor signal is proposed as a main informative parameter, and with large gaps the estimation based on the distance between the signal extremums is more preferred. The results of computer modeling of the influence of the joint gap on the parameters of the magnetic signal qualitatively coincide with the results of field measurements made at inspection speeds up to 60 km/h.

자동차용 기계부품의 홀 가공 품질 검사를 위한 비전 시스템

This paper presents a vision system for machined hole quality inspection of mechanical parts in an automobile. Automobile parts have various shapes and holes created by press punches. However, if the press punch pin is broken, a hole is not created on the mechanical parts. This problem causes serious part quality defects. To solve this problem, we proposed a vision system that could easily and cheaply inspect the quality of holes in automotive machining parts. A software development environment was created to build an economical vision inspection system. Images were gathered using the Near-real-time method to overcome the low frame-per-second of inexpensive Complementary Metal Oxide Semiconductor (CMOS) webcams. Status of the hole was determined using template matching and distance between holes as a feature. The hardware required for vision inspection was designed so that it could be directly applied to the automotive part manufacturing process. When the proposed vision inspection system was tested by installing it in an automobile parts factory for 3 months, the system showed an inspection accuracy of at least 97.9%. This demonstrates the effectiveness of the proposed method with accuracy and speed of hole defect inspection of machined parts.

Unsupervised defect segmentation on denim fabric via local patch prediction and residual fusion

Deep learning-based defect inspection has gained popularity in recent years. The dataset requirements for the supervised learning-based method are currently high, but the types of defects are numerous and difficult to gather. This work proposes a local image reconstruction-based unsupervised fabric defect segmentation method to address this problem. Cyclic structures make up the normal portion of the fabric image, whereas the defects are anomalous and minor in comparison. As a result, the defect will be recreated as a normal texture utilizing the information from its surrounding areas, and the defect information will be preserved in the residual image. By masking the same area with various shapes, different reconstruction outcomes and residual images can be achieved. The signal of the defect will be amplified and the noise will be decreased due to the random distribution when the generated residual pictures are fused, which can effectively identify the defect from the noise and lower the false detection rate. On the denim fabric dataset, the proposed unsupervised method can achieve high precision fabric defect segmentation, with the defect detection rate and detection precision reaching at least 85% and 89%, respectively, with high efficiency (approximately 60 m/min inspection speed), outperforming other fabric defect segmentation methods.

An Improved Algorithm for Insulator and Defect Detection Based on YOLOv4

To further improve the accuracy and speed of UAV inspection of transmission line insulator defects, this paper proposes an insulator detection and defect identification algorithm based on YOLOv4, which is called DSMH-YOLOv4. In the feature extraction network of the YOLOv4 model, the improved algorithm improves the residual edges of the residual structure based on feature reuse and designs the backbone network D-CSPDarknet53, which greatly reduces the number of parameters and computation of the model. The SA-Net (Shuffle Attention Neural Networks) attention model is embedded in the feature fusion network to strengthen the attention of target features and improve the weight of the target. Multi-head output is added to the output layer to improve the ability of the model to recognize the small target of insulator damage. The experimental results show that the number of parameters of the improved algorithm model is only 25.98% of that of the original model, and the mAP (mean Average Precision) of the insulator and defect is increased from 92.44% to 96.14%, which provides an effective way for the implementation of edge end algorithm deployment.