15 August 2026, Volume 44 Issue 4
    

  • Select all
    |
    Review
  • ZHANG Jianwei, ZHANG Guangxin, ZHOU Huina, LI Xuan, LIU Jinxiao, YANG Yingwei
    Physics Examination and Testing. 2026, 44(4): 1-5. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260021
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Non-destructive testing (NDT) is a core technology for ensuring the quality and safety of industrial products, key components, and major equipment. The standardization level of the personnel certification system directly determines the professional capabilities of practitioners and profoundly affects the reliability of Chinese manufacturing products and their international competitiveness. This paper reviews the development history of NDT personnel certification systems at home and abroad, and systematically analyzes the industry pattern, existing shortcomings, and international advanced experiences of Chinese certification system. Combined with the manufacturing power strategy, the "Belt and Road" initiative, the demand for high-quality manufacturing development and international production capacity cooperation, it extracts core development directions such as standardization construction, international mutual recognition, digital transformation, and industry resource integration, and proposes feasible optimization paths. The research can provide theoretical references and practical basis for the unification, standardization, internationalization, and high-quality development of Chinese NDT personnel certification system.
  • CHEN Changhua, SHI Feiyang, ZHANG Jianwei, FAN Hong
    Physics Examination and Testing. 2026, 44(4): 6-15. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260013
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Intelligent non-destructive testing grading technology is the core direction of technological advancement in the industrial non-destructive testing field. Leveraging the deep integration of artificial intelligence (AI), industrial robots, digital twins, and network communication technologies, it has established a stepwise and systematic development framework, evolving from manual inspection 1.0 to perceptual inspection 4.0. This advancement achieves a leapfrog upgrade in visual inspection intelligence, transitioning from "passive execution" to "active optimization," and from "human operating equipment" to "human commanding robots to operate equipment." It provides critical technical support for enhancing industrial inspection accuracy, optimizing efficiency, and adapting to complex working conditions. This article systematically elaborates on the technical connotations and application characteristics of each inspection level, with applicable scenarios covering various inspection needs in the industrial sector. These include perceptual inspection scenarios centered on artificial labor, automated equipment, AI technology, and human-machine collaboration; instrument operation scenarios characterized by manual operation, computer control, AI self-learning, and robotic execution; as well as non-destructive testing operations aimed at precision improvement, efficiency optimization, complex scenario adaptation, and flexible human-machine collaboration. In the future, intelligent robots and perceptual robots will form a clearly defined and complementary collaborative framework in product inspection, with manual inspection serving only as a supplementary method in extremely rare special cases. Perceptual robots will be deeply integrated into the entire production process, enabling efficient human-machine interaction and collaborative operations, thereby establishing a new working model where humans lead, machines collaborate, and intelligence empowers.
  • LIU Yanbo, LIU Yixin, LI Hongwei, LIU Jisheng, LI Min
    Physics Examination and Testing. 2026, 44(4): 16-31. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260047
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Air-coupled ultrasonic technology is a non-destructive testing technology that uses air as the coupling medium to achieve non-contact excitation, propagation, and signal acquisition of ultrasonic waves between the transducer, air, and the tested material, thereby enabling material evaluation. This technology does not require the use of liquid coupling agents, which can effectively reduce the risk of coupling agents contaminating or damaging sensitive materials in traditional ultrasonic testing. Therefore, it is conducive to integration with automated detection platforms and enables online detection of the internal quality of materials. Based on an exposition of the working principle and typical testing methods of air-coupled ultrasonic technology, this paper systematically reviews the core hardware and software systems of the technology, summarizes its typical applications in high-end manufacturing industries such as aerospace, new energy storage, and rail transit, and further analyzes its in-depth integration with artificial intelligence, robotics, and 5G communication technologies as well as its future development trends, demonstrating that this technology holds broad prospects for industrial applications.
  • LÜ Yingjie, ZHAO Boting, HU Qinlong
    Physics Examination and Testing. 2026, 44(4): 32-41. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260029
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Aviation composite materials, with their advantages of lightweight, high specific strength, high specific modulus, and fatigue resistance, have become core structural materials for weight reduction, efficiency improvement, and performance enhancement in new-generation civil and military aircraft. However, composite materials are prone to defects such as delamination, porosity, fiber folds, and debonding during their preparation and service processes, posing serious threats to flight safety. Ultrasonic nondestructive testing, characterized by strong penetrability, high sensitivity, intuitive imaging, and wide applicability, has become the mainstream technology for defect detection and quality evaluation of aviation composite materials. This paper systematically reviews the typical defect types and acoustic characteristics of aviation composite materials, summarizes the principles and research progress of mainstream technologies such as traditional ultrasonic testing, phased array ultrasound, air-coupled ultrasound, laser ultrasound, and ultrasonic guided waves, and compares the advantages and applicable scenarios of each technology. It focuses on advanced imaging algorithms such as full-focus imaging, 3D visualization, and tomographic imaging, as well as the application of artificial intelligence in defect identification and quantitative characterization. It summarizes the current challenges faced by these technologies in terms of adaptability to anisotropic materials, complex structure inspection, quantification of micro-defects, and high-temperature/in-service monitoring, and looks forward to future directions such as multimodal fusion, intelligent perception, digital twins, and standardized evaluation. The aim is to provide a comprehensive reference for the engineering application and innovative research and development of ultrasonic nondestructive testing technology for aviation composite materials.
  • Ultrasonic technique
  • ZHANG Yongchen, CHANG Xiaodong, WANG Hongliang, LI Changrong GUO Zemin, LI Guojin
    Physics Examination and Testing. 2026, 44(4): 42-50. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260042
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Aiming at the detection requirements of key components with complex curved surfaces in high-end equipment, and to solve the problems of poor adaptability, numerous scanning blind zones and inflexible path planning of traditional multi-axis C-scan (equal-depth scanning)equipment, an 8-axis linkage robotic arm ultrasonic testing system for complex curved surfaces is developed. The development of three core modules, including hardware system integration, independent C-scan software development and ultrasonic detection algorithm optimization, has been completed, achieving accurate detection of key components with complex curved surfaces. Key breakthroughs are achieved in core technologies including electromechanical cooperative control of multi-axis linkage systems, full-region path planning and profiling scanning technology for complex curved surfaces, autonomous ultrasonic C-scan imaging and control technology, and ultrasonic signal processing algorithms The test results indicate that the system has a complete and reliable hardware structure, and its human-computer interaction and software interface are reasonable and user-friendly. The optimized algorithm is reliable and verified by experimental data to improve the signal-to-noise ratio from 4.4 dB to 18 dB. The system can perform a full scan of all 9 surfaces of complex disk and ring components without any blind zones. It has a stable detection capability for ϕ0.4 mm and ϕ0.8 mm micro flat-bottom holes and transverse holes, with positioning errors less than 1% for both. The detection results have good consistency and high stability. The system has been successfully applied in engineering for internal testing scenarios, realizing the domestic substitution and industrialization of ultrasonic testing equipment for complex components of high-end equipment.
  • SHI Qiang, REN Keshe, XU Wei, XU Xin, LIN Min, CHEN Changhua
    Physics Examination and Testing. 2026, 44(4): 51-60. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260011
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Internal defects in high-quality large-section continuous casting billets are a primary cause of quality disputes in special steel production. To address the issues of strong subjectivity and susceptibility to missed detections and misclassifications in handheld ultrasonic flaw detection, this paper introduces machine learning methods to achieve automatic classification and identification of internal defects in large-section 42CrMoA continuous casting billets through time-domain feature extraction and modeling of ultrasonic A-scan signals. A total of 13 time-domain features are extracted, and classifiers including fine-grained decision tree and support vector machine (SVM) as well as a feedforward neural network are constructed. The neural network is trained using three algorithms: Levenberg-Marquardt (LM), Bayesian regularization (BR), and scaled conjugate gradient (SCG). The results show that SVM achieves the highest classification accuracy of 99.8%. Among the three neural network algorithms, BR exhibits the best fitting accuracy, while LM and SCG show faster convergence, with SCG striking a balance between convergence speed and generalization performance. Regression analysis indicates that the correlation coefficient between the output and target values of the SCG model exceeds 0.98, validating its effectiveness. The study demonstrates that machine learning methods can effectively replace manual judgment for intelligent identification of internal defects in billets. Based on the above comparative analysis, this paper proposes a cascading strategy of “preliminary screening with traditional methods followed by refined inspection with neural networks”, providing a feasible technical pathway for intelligent quality assessment of continuous casting billets in special steel enterprises.
  • WANG Hongliang, QIN Zheng, LI Changrong, ZHANG Yongchen GUO Zemin, CHANG Xiaodong, ZHANG Shangyi, ZHANG Hailong
    Physics Examination and Testing. 2026, 44(4): 61-71. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260045
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Large-diameter steel bars are essential structural materials in engineering applications, and the existence of their internal defects will directly affect the service safety of components. To address the current challenges of low inspection efficiency and insufficient automation in ultrasonic testing of large-diameter steel bars, a fully water immersion phased array ultrasonic testing (PAUT) automatic system has been developed. The system integrates electronic focusing of phased array technology with C-scan imaging and employs a spiral scanning method to achieve full-volume coverage for steel bars with diameters ranging from ϕ100 mm to ϕ355 mm. Full automation of the inspection process is realized through integration of automatic loading and unloading, intelligent transfer units, and a programmable logic controller (PLC) system. Using artificial defects reference bar specimens containing flat-bottom holes and transverse holes, the tests on key indicators such as signal-to-noise ratio, stability and blind zone are conducted according to YB/T 4082-2020 standard. The performance evaluations are compared with multi-channel fully water immersion ultrasonic systems and through-type phased array systems.The results show that the developed system can accurately detect the specified equivalent artificial defects. Specifically, it achieves a signal-to-noise ratio (SNR) not less than 12 dB, the detection sensitivity of the same channel with of a fluctuation not more than 1 dB over a continuous 4 h period, and an end blind detection zone not greater than 10 mm. All these three indicators are higher than the requirements of YB/T 4082-2020 standard. Under the premise of maintaining equivalent inspection sensitivity, its inspection efficiency is significantly higher than that of the multi-channel full water immersion ultrasonic testing method. Moreover, it effectively solves the technical limitation that the existing through-type phased array systems are difficult to inspect large-diameter bars with diameter over ϕ220 mm. The system has been successfully deployed in an industrial production line, which has verified its engineering applicability and reliability, providing an effective non-destructive testing solution for quality assurance of large-diameter steel bars.
  • SUN Shaoguang, ZHOU Bo, XU Lei, WANG Qing, XIE Guangqun, LÜ Dan
    Physics Examination and Testing. 2026, 44(4): 72-79. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260018
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In response to the difficulties in implementing online automatic detection for nuclear power steel plates, the low efficiency of traditional manual ultrasonic testing, and the high labor intensity, this paper designs an automatic ultrasonic detection device for steel plates that can simulate the manual ultrasonic detection process. The device is composed of an omnidirectional travelling flaw detection trolley, a laser positioning guidance mechanism, a multi-functional probe holder, a multi-channel ultrasonic detector, a power supply and coupling water supply unit. It uses servo motors to drive McNamara wheels to achieve omnidirectional movement. Combined with laser guidance and multi-channel detection systems, it completes the full-automatic scanning of the plate surface along the preset trajectory. It is suitable for rapid and automatic detection of steel plates with a thickness of 20-100 mm. Field tests show that the device can effectively detect internal layered and non-layered defects, with detection rate and determination accuracy comparable to manual detection; the detection efficiency is three times of traditional manual detection, and the false alarm rate of defects can be controlled within 3%, which can meet the personalized and high-standard non-destructive testing requirements for high grade steel plates.
  • ZHUANG Ya, ZHOU Zhiwei, JI Bing, WANG Yifei, FANG Kun, KANG Xiangyang
    Physics Examination and Testing. 2026, 44(4): 80-89. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260034
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Multilayer foam bonding composites contains a large number of porous/honeycomb structures, making it difficult to achieve effective ultrasonic coupling and incidence when conducting non-destructive testing with conventional ultrasonic techniques. This results in ultrasonic detection of internal defects becoming a common industry challenge. This paper employs air as the ultrasonic coupling medium for non-destructive testing of multilayer foam bonding composites, and determines key process parameters based on measured data from this type of composite. The key detection parameters of air-coupled ultrasonic technique (ACUT), such as probe spacing, pulse string number, average count, center frequency, preamplification ratio, and filtering frequency, are selected as the research variables. The influence laws of each parameter on the amplitude of defect echo signals are discussed, and the optimized parameters suitable for ultrasonic testing of multi-layer foam bonding composites are determined: the probe spacing is 250 mm, the pulse string number is 5, the average count is 1, the center frequency is 140 kHz, the preamplification ratio is 20%, and the filtering frequency is 80-200 Hz. By preparing multilayer foam bonding composite defect samples with different depths and sizes at the interface of functional materials, the air-coupled ultrasonic technology with optimized parameters is used for detection. The results show that the maximum relative errors of detection results are 7.50%, 5.00% and 4.00% for prefabricated circular defects with diameters of 20, 30 and 50 mm, respectively. This confirms that the air-coupled ultrasonic technique can achieve quantitative detection of interfacial debonding defects at different depths inside this type of composite.
  • XU Hong, ZHU Guomiao, WANG Liang, ZHOU Haiyan, GUO Tao
    Physics Examination and Testing. 2026, 44(4): 90-95. https://doi.org/10.13228/j.boyuan.issn1001-0777.20250107
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    When inspecting welded joints of P91 steel pipelines in power plant boilers, the surface temperature ranges from 50 ℃ to 300 ℃. This temperature is higher than 0-50 ℃ that is specified in the current standard of NB/T 47013.3-2023, making it a major challenge for periodic inspection of embedded defects in boilers. This paper fabricated an equivalent calibration test block using P91 steel and designed a special detection probe for high-temperature measurement. The probe wedge was made of polyimide. Then, a system was established for determinating the ultrasonic shear wave velocity of P91 steel within the temperature range of 50 to 300 ℃combined with 900 ℃ high-temperature coupling agent. This solved the problem of large positioning error in high-temperature ultrasonic testing of P91 steel welding joints. The results showed that in the temperature range of 50-300 ℃, the ultrasonic shear wave velocity of P91 steel decreased gradually as temperature rised. At the same temperature, the ultrasonic shear wave velocity of P91 steel was higher than that of 20# steel. Errors generated during scanning baseline calibration and ultrasonic probe fabrication were non-negligible. Therefore, it is necessary to use P91 steel to fabricate an equivalent calibration test block to calibrate the scanning baseline and to manufacture the ultrasonic probe.
  • ZHOU Haohao, DING Jie, MA Jun, HUANG Yin
    Physics Examination and Testing. 2026, 44(4): 96-102. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260022
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The national standard of NB/T 47013.3-2023 (hereinafter referred to as the 2023 edition) has been released, and the performance requirements for A-scan pulse-echo ultrasonic flaw detectors are mainly concentrated in Appendix A. In the actual application process, Appendix A only clearly defines the limit values of various performance indicators. The regulations regarding technical details such as specific test conditions, operation steps and other technical details are more concise compared to other calibration standards for ultrasonic equipment, making it difficult for third-party testing institutions to establish a complete testing capabilities for ultrasonic flaw detectors based on this appendix. On the other hand, there is a widespread demand among enterprises engaged in the special inspection of pressure equipment for ultrasonic flaw detector testing reports based on Appendix A of 2023 edition standard. Therefore, the research on practical testing methods for the parameters listed in Appendix A holds significant value for engineering applications. This paper conducts a comparative analysis with the internationally universal ultrasonic flaw detector testing standard of ISO 22232-1:2020, and investigates the testing implementation methods for each parameter. The results show that the test methods specified in standard of ISO 22232-1:2020 are more systematic and detailed. Through experimental verification of selected parameters such as transmitter pulse voltage, dead time after transmit pulse, dynamic range and the assessment of measurement uncertainty, the feasibility and validity of adopting the test method specified in standard of ISO 22232-1:2020 to replace Appendix A of 2023 edition standard for testing verification are validated.
  • WANG Yingfeng, WANG Yongfeng, YANG Jingwei, ZHANG Qing, TAN Ning, YANG Bin
    Physics Examination and Testing. 2026, 44(4): 103-106. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260020
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Ultrasonic testing technology, with critically refracted longitudinal wave method as the core, is an important means for the nondestructive testing of residual stress in materials. Based on the acoustoelastic effect, this paper establishes a mathematical model for ultrasonic residual stress measurement, derives the calculation formula of measurement uncertainty, and systematically identifies the main sources of uncertainty throughout the whole testing process. In accordance with the specification JJF 1059.1-2012 of Evaluation and Expression of Uncertainty in Measurement, type A and type B uncertainty components are evaluated for key influencing factors including repeatability measurement, acoustoelastic coefficient, time-of-flight resolution of instrument, ambient temperature, and material differences. The combined standard uncertainty and expanded uncertainty are calculated, and a complete expression method for testing results is formed. The evaluation system proposed in this paper is both standardized and operable, which can provide a unified guideline for the reliability evaluation of laboratory and on-site ultrasonic residual stress testing results, and promote the standardization and large-scale application of this technology in the field of engineering inspection.
  • Electromagnetic, radiographic, and optical technologies
  • YUAN Xin’an, WANG Xiangyang, JIA Shimin, QI Changchao LIN Dong, ZHANG Miao, LI Wei, XIN Jingyu
    Physics Examination and Testing. 2026, 44(4): 107-117. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260017
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Long-distance oil and gas pipelines are subjected to the coupling effect of multiple factors such as geological disasters and pressure fluctuation of transported medium, which easily causes local abrupt stress variation. Pipeline stress inspection can effectively prevent pipeline failure and reduce or avoid the occurrence of safety accidents. Based on the basic theory of alternating current stress measurement (ACSM), this paper establishes the ACSM simulation model and explores the variation law of electromagnetic parameters under stress. Consequently, an alternating current field measurement(ACFM) probe applicable to pipeline stress detection is developed, and a pipeline internal inspection system is integrated. Static tensile tests are carried out on common pipeline steels including Q235, X60, X70 and X80. Experimental results show that as the increase of tensile stress, the magnetic flux density (Bx) signal shows a monotonically increasing trend along the direction of the applied magnetic field. Under different probe excitation parameters, the sensitivity of the Bx signal varies. With the increase of excitation frequency, the sensitivity of the Bx signal first increases and then decreases. The Bx signal is adopted to characterize stress. A mapping database between Bx signal and stress sensitivity is established for common pipeline steels. Combined with actual pipeline operating conditions and signal variation characteristics, a combined denoising method of amplitude limiting and smoothing is applied to realize signal noise reduction. Meanwhile, the quantitative evaluation of stress in abnormal regions is achieved, which provides a reference for the quantitative stress detection of long-distance oil and gas pipelines. Field tests are carried out on X80 pipeline steel pipelines with natural gas as the transported medium, then the 8-array probe and the stress detection electronic unit are mounted on a crawling device. The self-developed detection system is adopted to extract and analyze the measured data, and a total of 4 obvious stress abnormal regions are identified. This study verifies the feasibility of applying ACFM technology to quantitative pipeline stress detection, and provides reliable technical support for the safety early warning of long-distance pipelines.
  • SUN Dandan, DONG Caichang
    Physics Examination and Testing. 2026, 44(4): 118-126. https://doi.org/10.13228/j.boyuan.issn1001-0777.20250094
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Copper-nickel alloys are widely adopted in seawater pipeline system of ships, which plays a crucial role in ensuring the normal operation of marine power plants, auxiliary machinery, and related equipment. To investigate the differences in corrosion behavior of copper-nickel alloy pipes from different sources, this paper developed an in-situ characterization method for corrosion behavior based on white light interferometry(WLI). This method gradually improved the measurement accuracy through steps such as screening test parameters and unifying the reference plane. The measurement results were verified by the converted values of weight loss method and the results btained by focused ion beam(FIB) cross-section method, ultimately realizing the precise measurement of corrosion morphology, corrosion thickness loss, and corrosion product film thickness. By applying this method to analyze the corrosion behavior of B10 copper-nickel alloy pipes, the main factors affecting the corrosion resistance of three different types of B10 copper-nickel alloy were identified. The results showed that the in-situ characterization method based on FIB established in this study could accurately characterize corrosion heights of 0.2 μm and above. Combined with characterization methods like FIB, it provides a new approach for evaluating the corrosion behavior of B10 copper-nickel alloy pipes.
  • HOU Shengli, DING Xuelong
    Physics Examination and Testing. 2026, 44(4): 127-132. https://doi.org/10.13228/j.boyuan.issn1001-0777.20250043
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In the fluorescent penetrant testing (FPT) of surface defects on aerospace parts, the fluorescent indications are normally observed and evaluated by visual inspection or magnifying glass, then measured and graded with fluorescent comparison scale. However, observation with magnifying glass is limited by insufficient resolution, and it could not realize the recording and preservation of fluorescent images; the fluorescent comparison scale could only provide the size range of fluorescent indications, with obvious limitations in measurement accuracy. Based on the imaging principle of biological fluorescence microscopes, this paper optimizes and reforms the optical path system of a metallographic stereomicroscope. By integrating high-brightness light-emitting diode (LED) ultraviolet light sources and installing additional filtering devices at the light source, objective lens and camera interface, an ultraviolet/white light dual-source microscope is successfully developed. This microscope has enabled the simultaneous observation and collection of fluorescence images and white light images of in weld seam crater and cavitation defects on the inner bore of hydraulic housing, as well as the regular testing of standard crack specimens. It could precisely measure the multi-dimensional dimensions of the defects with measurement accuracy of ±0.01 mm, which effectively solves the technical problems in qualitative evaluation and quantitative measurement of fluorescent penetrant testing.
  • LI Song
    Physics Examination and Testing. 2026, 44(4): 133-140. https://doi.org/10.13228/j.boyuan.issn1001-0777.20260006
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Aiming at the problems of weak defect features and complex background noise in digital radiography (DR) detection images of pressure vessel welds, traditional methods adopt gray histogram and global thresholding to implement defect segmentation and identification. As a single gray feature-based image processing method without integrating geometric and texture information, it suffers from low recognition accuracy and poor robustness. This paper presents an intelligent weld defect recognition approach based on the Gradient Boosting Tree (GBT) model. Without relying on complex deep networks, this method takes multi-dimensional features including gray, texture and geometry as inputs, and constructs a classification model suitable for weld defects through adaptive feature weight assignment and iterative fitting optimization. Meanwhile, a combined strategy of adaptive threshold segmentation, directional filtering and wavelet transform is adopted to achieve accurate extraction of defect features. Experimental results demonstrate that the proposed method, namely “gray + texture + geometry fusion + GBT model”, achieves 91.2% accuracy for crack defect recognition, 84.3% for porosity defect recognition, 91.0% for incomplete penetration defect recognition and 90.0% for lack of fusion defect recognition. The overall classification performance of the four key defects is significantly superior to that of traditional detection methods using only single gray feature. The model has the advantages of small parameters, fast inference speed and low hardware requirements, stable reliablility under small sample and complex working conditions, which is suitable for lightweight deployment in industrial fields. This method can effectively improve the detection efficiency and recognition accuracy of pressure vessel weld defects, and provide lightweight technical support for the intellectualization of industrial non-destructive testing.