期刊文献+
共找到16,075篇文章
< 1 2 250 >
每页显示 20 50 100
A self-sensing HTPB liner for non-destructive monitoring nitroglycerin(NG)migration at the interface between double base propellant and the HTPB liner
1
作者 Jie Wang Bo Liu +4 位作者 Yanchun Li Mengqi Chen Qian Guo Dongming Song Aifeng Jiang 《Defence Technology(防务技术)》 2025年第8期166-175,共10页
During the storage of composite propellants, the migration of plasticizers and other unbonded additives at the interfaces of liner adhesives has garnered significant attention in understanding liner failure mechanisms... During the storage of composite propellants, the migration of plasticizers and other unbonded additives at the interfaces of liner adhesives has garnered significant attention in understanding liner failure mechanisms, aging processes, and safety performance. However, there is currently no non-destructive and quantitative detection method for migration of plasticizers in propellant liner. In this study, we developed a HTPB sensing liner by incorporating conductive fillers-namely carbon black(CB), carbon nanotubes(CNTs), and graphene nanoplatelets(GNP)-into the HTPB matrix. The synergistic interaction between CNTs and GNP facilitates the formation of a tunneling conductive network that imparts electrical conductivity to the HTPB liner. To elucidate the functional relationship between conductivity and nitroglycerin(NG) migration, we applied the HTPB sensing liner onto double base propellant surfaces and measured both the conductivity of the sensing layer and NG migration during a 71°C accelerated aging experiment. The results shows that when CNTs/GNP content reaches 3wt%, there is an exponential correlation between conductivity and NG migration with a fitting degree of 0.9652;the average response sensitivity of ΔR/R0 relative to NG migration is calculated as 41.69, with an average deviation of merely5.67% between NG migrations derived from conductivity fittings compared to those obtained via TGA testing results. Overall, this sensing liner exhibits excellent capabilities for detecting NG migration nondestructively and quantitatively while offering a novel approach for assessing interfacial component migrations as well as debonding defects in propellants-a promising avenue for future self-monitoring strategies regarding propellant integrity. 展开更多
关键词 Sensing liner Electrical conductivity Nitroglycerine migration non-destructive detection
在线阅读 下载PDF
Status and Development of Rapid Detection Technology for Tunnel Structural Defects 被引量:3
2
作者 LIU Xuezeng FANG Maoliu +3 位作者 WU Dexing LI Yinping LIU Xingen LI Gang 《隧道建设(中英文)》 北大核心 2025年第4期657-676,I0005-I0024,共40页
Based on inspection data,the authors analyze and summarize the main types and distribution characteristics of tunnel structural defects.These defects are classified into three types:surface defects,internal defects,an... Based on inspection data,the authors analyze and summarize the main types and distribution characteristics of tunnel structural defects.These defects are classified into three types:surface defects,internal defects,and defects behind the structure.To address the need for rapid detection of different defect types,the current state of rapid detection technologies and equipment,both domestically and internationally,is systematically reviewed.The research reveals that surface defect detection technologies and equipment have developed rapidly in recent years.Notably,the integration of machine vision and laser scanning technologies have significantly improved detection efficiency and accuracy,achieving crack detection precision of up to 0.1 mm.However,the non-contact rapid detection of internal and behind-the-structure defects remains constrained by hardware limitations,with traditional detection remaining dominant.Nevertheless,phased array radar,ultrasonic,and acoustic vibration detection technologies have become research hotspots in recent years,offering promising directions for detecting these challenging defect types.Additionally,the application of multisensor fusion technology in rapid detection equipment has further enhanced detection capabilities.Devices such as cameras,3D laser scanners,infrared thermal imagers,and radar demonstrate significant advantages in rapid detection.Future research in tunnel inspection should prioritize breakthroughs in rapid detection technologies for internal and behind-the-structure defects.Efforts should also focus on developing multifunctional integrated detection vehicles that can simultaneously inspect both surface and internal structures.Furthermore,progress in fully automated,intelligent systems with precise defect identification and real-time reporting will be essential to significantly improve the efficiency and accuracy of tunnel inspection. 展开更多
关键词 TUNNEL structural defect inspection techniques inspection equipment rapid inspection
在线阅读 下载PDF
Defect Dipole Thermal-stability to the Electro-mechanical Properties of Fe Doped PZT Ceramics
3
作者 SUN Yuxuan WANG Zheng +5 位作者 SHI Xue SHI Ying DU Wentong MAN Zhenyong ZHENG Liaoying LI Guorong 《无机材料学报》 北大核心 2025年第5期545-551,I0009-I0010,共9页
The accepted doping ion in Ti^(4+)-site of PbZr_(y)Ti_(1–y)O_(3)(PZT)-based piezoelectric ceramics is a well-known method to increase mechanical quality factor(Q_(m)),since the acceptor coupled by oxygen vacancy beco... The accepted doping ion in Ti^(4+)-site of PbZr_(y)Ti_(1–y)O_(3)(PZT)-based piezoelectric ceramics is a well-known method to increase mechanical quality factor(Q_(m)),since the acceptor coupled by oxygen vacancy becomes defect dipole,which prevents the domain rotation.In this field,a serious problem is that generally,Qm decreases as the temperature(T)increases,since the oxygen vacancies are decoupled from the defect dipoles.In this work,Q_(m) of Pb_(0.95)Sr_(0.05)(Zr_(0.53)Ti_(0.47))O_(3)(PSZT)ceramics doped by 0.40%Fe_(2)O_(3)(in mole)abnormally increases as T increases,of which the Qm and piezoelectric coefficient(d_(33))at room temperature and Curie temperature(TC)are 507,292 pC/N,and 345℃,respectively.The maximum Qm of 824 was achieved in the range of 120–160℃,which is 62.52%higher than that at room temperature,while the dynamic piezoelectric constant(d_(31))was just slightly decreased by 3.85%.X-ray diffraction(XRD)and piezoresponse force microscopy results show that the interplanar spacing and the fine domains form as temperature increases,and the thermally stimulated depolarization current shows that the defect dipoles are stable even the temperature up to 240℃.It can be deduced that the aggregation of oxygen vacancies near the fine domains and defect dipole can be stable up to 240℃,which pins domain rotation,resulting in the enhanced Q_(m) with the increasing temperature.These results give a potential path to design high Q_(m) at high temperature. 展开更多
关键词 defect dipole temperature characteristic oxygen vacancy electro-mechanical property mechanical quality factor hardening doping
在线阅读 下载PDF
Defect engineering of carbon-based electrocatalysts for the CO_(2)reduction reaction:A review 被引量:2
4
作者 LU Yan-kun CHENG Bai-xue +1 位作者 ZHAN Hao-yu ZHOU Peng 《新型炭材料(中英文)》 SCIE EI CAS CSCD 北大核心 2024年第1期17-41,共25页
Electrocatalytic carbon dioxide(CO_(2))reduction is an important way to achieve carbon neutrality by converting CO_(2)in-to high-value-added chemicals using electric energy.Carbon-based materials are widely used in va... Electrocatalytic carbon dioxide(CO_(2))reduction is an important way to achieve carbon neutrality by converting CO_(2)in-to high-value-added chemicals using electric energy.Carbon-based materials are widely used in various electrochemical reactions,including electrocatalytic CO_(2)reduction,due to their low cost and high activity.In recent years,defect engineering has attracted wide attention by constructing asymmetric defect centers in the materials,which can optimize the physicochemical properties of the mater-ial and improve its electrocatalytic activity.This review summarizes the types,methods of formation and defect characterization tech-niques of defective carbon-based materials.The advantages of defect engineering and the advantages and disadvantages of various defect formation methods and characterization techniques are also evaluated.Finally,the challenges of using defective carbon-based materials in electrocatalytic CO_(2)reduction are investigated and opportunities for their use are discussed.It is believed that this re-view will provide suggestions and guidance for developing defective carbon-based materials for CO_(2)reduction. 展开更多
关键词 defect engineering Carbon-based materials ELECTROCATALYSIS CO_(2)reduction
在线阅读 下载PDF
Research on Transfer Learning in Surface Defect Detection of Printed Products 被引量:1
5
作者 ZHU Xin-yu SI Zhan-jun CHEN Zhi-yu 《印刷与数字媒体技术研究》 CAS 北大核心 2024年第6期38-44,共7页
To advance the printing manufacturing industry towards intelligence and address the challenges faced by supervised learning,such as the high workload,cost,poor generalization,and labeling issues,an unsupervised and tr... To advance the printing manufacturing industry towards intelligence and address the challenges faced by supervised learning,such as the high workload,cost,poor generalization,and labeling issues,an unsupervised and transfer learning-based method for printing defect detection was proposed in this study.This method enabled defect detection in printed surface without the need for extensive labeled defect.The ResNet101-SSTU model was used in this study.On the public dataset of printing defect images,the ResNet101-SSTU model not only achieves comparable performance and speed to mainstream supervised learning detection models but also successfully addresses some of the detection challenges encountered in supervised learning.The proposed ResNet101-SSTU model effectively eliminates the need for extensive defect samples and labeled data in training,providing an efficient solution for quality inspection in the printing industry. 展开更多
关键词 Transfer learning UNSUPERVISED defect detection PRINTING
在线阅读 下载PDF
Microscopic defects formation and dynamic mechanical response analysis of Q345 steel plate subjected to explosive load 被引量:1
6
作者 Zhengqing Zhou Zechen Du +6 位作者 Yulong Zhang Guili Yang Ruixiang Wang Yuzhe Liu Peize Zhang Yaxin Zhang Xiao Wang 《Defence Technology(防务技术)》 SCIE EI CAS CSCD 2024年第2期430-442,共13页
As the basic protective element, steel plate had attracted world-wide attention because of frequent threats of explosive loads. This paper reports the relationships between microscopic defects of Q345 steel plate unde... As the basic protective element, steel plate had attracted world-wide attention because of frequent threats of explosive loads. This paper reports the relationships between microscopic defects of Q345 steel plate under the explosive load and its macroscopic dynamics simulation. Firstly, the defect characteristics of the steel plate were investigated by stereoscopic microscope(SM) and scanning electron microscope(SEM). At the macroscopic level, the defect was the formation of cave which was concentrated in the range of 0-3.0 cm from the explosion center, while at the microscopic level, the cavity and void formation were the typical damage characteristics. It also explains that the difference in defect morphology at different positions was the combining results of high temperature and high pressure. Secondly, the variation rules of mechanical properties of steel plate under explosive load were studied. The Arbitrary Lagrange-Euler(ALE) algorithm and multi-material fluid-structure coupling method were used to simulate the explosion process of steel plate. The accuracy of the method was verified by comparing the deformation of the simulation results with the experimental results, the pressure and stress at different positions on the surface of the steel plate were obtained. The simulation results indicated that the critical pressure causing the plate defects may be approximately 2.01 GPa. On this basis, it was found that the variation rules of surface pressure and microscopic defect area of the Q345 steel plate were strikingly similar, and the corresponding mathematical relationship between them was established. Compared with Monomolecular growth fitting models(MGFM) and Logistic fitting models(LFM), the relationship can be better expressed by cubic polynomial fitting model(CPFM). This paper illustrated that the explosive defect characteristics of metal plate at the microscopic level can be explored by analyzing its macroscopic dynamic mechanical response. 展开更多
关键词 Explosive load Q345 steel Micro defect Finite element simulation Dynamic response Data fitting
在线阅读 下载PDF
Molecular Dynamics Simulation of Shock Response of CL-20 Co-crystals Containing Void Defects 被引量:1
7
作者 Changlin Li Wei Yang +5 位作者 Qiang Gan Yajun Wang Lin Liang Wenbo Zhang Shuangfei Zhu Changgen Feng 《Defence Technology(防务技术)》 SCIE EI CAS CSCD 2024年第1期364-374,共11页
To investigate the effect of void defects on the shock response of hexanitrohexaazaisowurtzitane(CL-20)co-crystals,shock responses of CL-20 co-crystals with energetic materials ligands trinitrotoluene(TNT),1,3-dinitro... To investigate the effect of void defects on the shock response of hexanitrohexaazaisowurtzitane(CL-20)co-crystals,shock responses of CL-20 co-crystals with energetic materials ligands trinitrotoluene(TNT),1,3-dinitrobenzene(DNB),solvents ligands dimethyl carbonate(DMC) and gamma-butyrolactone(GBL)with void were simulated,using molecular dynamics method and reactive force field.It is found that the CL-20 co-crystals with void defects will form hot spots when impacted,significantly affecting the decomposition of molecules around the void.The degree of molecular fragmentation is relatively low under the reflection velocity of 2 km/s,and the main reactions are the formation of dimer and the shedding of nitro groups.The existence of voids reduces the safety of CL-20 co-crystals,which induced the sensitivity of energetic co-crystals CL-20/TNT and CL-20/DNB to increase more significantly.Detonation has occurred under the reflection velocity of 4 km/s,energetic co-crystals are easier to polymerize than solvent co-crystals,and are not obviously affected by voids.The results show that the energy of the wave decreases after sweeping over the void,which reduces the chemical reaction frequency downstream of the void and affects the detonation performance,especially the solvent co-crystals. 展开更多
关键词 CL-20 co-crystals Molecular dynamics simulation Reactive forcefield Impact response Hot spot Void defect
在线阅读 下载PDF
Micro defects formation and dynamic response analysis of steel plate of quasi-cracking area subjected to explosive load
8
作者 Zheng-qing Zhou Ze-chen Du +5 位作者 Xiao Wang Hui-ling Jiang Qiang Zhou Yu-long Zhang Yu-zhe Liu Pei-ze Zhang 《Defence Technology(防务技术)》 SCIE EI CAS CSCD 2024年第1期580-593,共14页
As the protective component,steel plate had attracted extensive attention because of frequently threats of explosive loads.In this paper,the evolution of microstructure and the mechanism of damage in the quasi-crackin... As the protective component,steel plate had attracted extensive attention because of frequently threats of explosive loads.In this paper,the evolution of microstructure and the mechanism of damage in the quasi-cracking area of steel plate subjected to explosive load were discussed and the relationships between micro defects and dynamic mechanical response were revealed.After the explosion experiment,five observation points were selected equidistant from the quasi-cracking area of the section of the steel plate along the thickness direction,and the characteristics of micro defects at the observation points were analyzed by optical microscope(OM),scanning electron microscope(SEM) and electron backscattered diffraction(EBSD).The observation result shows that many slip bands(SBs) appeared,and the grain orientation changed obviously in the steel plate,the two were the main damage types of micro defects.In addition,cracks,peeling pits,grooves and other lager micro defects were appeared in the lower area of the plate.The stress parameters of the observation points were obtained through an effective numerical model.The mechanism of damage generation and crack propagation in the quasicracking area were clarified by comparing the specific impulse of each observation point with the corresponding micro defects.The result shows that the generation and expansion of micro defects are related to the stress area(i.e.the upper compression area,the neutral plane area,and the lower tension area).The micro defects gather and expand at the grain boundary,and will become macroscopic damage under the continuous action of tensile stress.Besides,the micro defects at the midpoint of the section of the steel plate in the direction away from the explosion center(i.e.the horizontal direction) were also studied.It was found that the specific impulse at these positions were much smaller than that in the thickness direction,the micro defects were only SBs and a few micro cracks,and the those decreased with the increase of the distance from the explosion center. 展开更多
关键词 Explosive load Quasi-cracking area Micro defects Steel plate Dynamic response Numerical simulation
在线阅读 下载PDF
基于无人机巡检的风机叶片表面缺陷检测技术 被引量:5
9
作者 谭兴国 张高明 《电测与仪表》 北大核心 2025年第3期183-189,共7页
“双碳”背景下大力发展新能源尤为重要,风力发电是重要的清洁能源,在新能源领域中风电规模也在一并扩大。随着风电机组的大型化,叶片受损概率也在增加,针对大型风力机叶片缺陷检测成本高、工作环境差等问题,文中提出了一种基于无人机... “双碳”背景下大力发展新能源尤为重要,风力发电是重要的清洁能源,在新能源领域中风电规模也在一并扩大。随着风电机组的大型化,叶片受损概率也在增加,针对大型风力机叶片缺陷检测成本高、工作环境差等问题,文中提出了一种基于无人机采集图像和数字图像处理的风机叶片表面缺陷检测方法。针对无人机采集图像的特点,应用加权平均值实现灰度处理,再应用中值滤波实现图像降噪;并提出限制对比度自适应直方图均衡化(contrast limited adaptive histogram equalization,CLAHE)算法对图像进行增强,使目标区域和缺陷处细节更加清晰完整,提升了检测效率;通过图像前景分割及阈值处理等分离提取缺陷的特征信息,并由连通域进行框取,实现叶片表面的检测。通过引入性能评价指标平均交并比(mean intersection over union,MI-oU)来计算检测缺陷图像的准确率与误差率,经实验验证所提方法对砂眼、划痕、裂纹等典型叶片缺陷的检测准确率均在90%以上,尤其是裂纹缺陷的检测准确率可达到95%,从而验证了文章检测方法的有效性。 展开更多
关键词 风电叶片 叶片缺陷 对比度自适应直方图均衡化 缺陷检测 无人机
在线阅读 下载PDF
含局部缺陷的角接触球轴承时变位移激励及动力学建模 被引量:2
10
作者 雷春丽 宋瑞哲 +3 位作者 樊高峰 刘凯 薛伟 李建华 《北京航空航天大学学报》 北大核心 2025年第3期753-762,共10页
角接触球轴承在长时间工作的情况下,会产生故障损伤从而影响系统正常运行。以外圈具有局部缺陷的角接触球轴承为研究对象,提出不同局部缺陷轮廓的判别方法,建立角接触球轴承局部缺陷时变位移激励广义表征模型,研究局部缺陷演化过程及其... 角接触球轴承在长时间工作的情况下,会产生故障损伤从而影响系统正常运行。以外圈具有局部缺陷的角接触球轴承为研究对象,提出不同局部缺陷轮廓的判别方法,建立角接触球轴承局部缺陷时变位移激励广义表征模型,研究局部缺陷演化过程及其位移激励机理。在此基础上,考虑轴承缺陷引起的时变位移对动力学特性的影响,基于赫兹接触理论,建立角接触球轴承故障动力学模型,并通过实验验证了所建模型的正确性。分析结果表明:矩形局部缺陷最终会演化成梯形形态;不同缺陷形貌诱发的位移激励变化趋势是不同的;与局部缺陷的长度相比,宽度对位移激励的影响更大。研究结果对轴承优化设计和故障诊断提供了理论基础。 展开更多
关键词 角接触球轴承 局部缺陷 时变位移激励 缺陷演变 动力学建模
在线阅读 下载PDF
基于改进SSD的螺栓缺陷检测算法 被引量:1
11
作者 宿景芳 杜金梦 +3 位作者 王震洲 于平平 王建超 张佳佳 《传感器与微系统》 北大核心 2025年第5期129-132,137,共5页
螺栓是输电线重要组件,针对航拍输电线巡检图像中螺栓尺寸小、背景复杂的问题,提出了一种基于改进SSD的螺栓缺陷检测算法。首先,将SSD骨干网络VGG-16替换为DenseNet并在其基础上,将GAM注意力机制融入DenseNet卷积层使模型更好地捕捉全... 螺栓是输电线重要组件,针对航拍输电线巡检图像中螺栓尺寸小、背景复杂的问题,提出了一种基于改进SSD的螺栓缺陷检测算法。首先,将SSD骨干网络VGG-16替换为DenseNet并在其基础上,将GAM注意力机制融入DenseNet卷积层使模型更好地捕捉全局信息;其次,将DenseNet过渡层的平均池化操作进行替换来加深网络层数,从而增强网络特征提取能力。此外,设计了一种多尺度特征层融合模块(IPANet),结合主干网络改进后的SSD网络模型,将网络结构中的底层视觉特征和高层语义特征有机结合,从而获得更多位置信息以解决螺栓尺度小的问题。最后,自主构建了航拍输电线螺栓数据集,并进行数据增强。实验结果表明:改进算法在自主构建数据集上mAP相比原SSD算法提升了2.79%,与其他主流算法比较,改进算法检测精度和检测速度均优于其他主流算法。 展开更多
关键词 缺陷检测 DenseNet 注意力机制
在线阅读 下载PDF
基于改进YOLOv5s太阳能电池片表面缺陷检测算法 被引量:2
12
作者 王巍 余欣 +1 位作者 缪佳欣 刘晓宇 《电子测量技术》 北大核心 2025年第5期128-136,共9页
太阳能电池片表面缺陷的出现会严重影响太阳能转化效率,准确检测太阳能电池片表面缺陷并及时处理可以有效提高发电效率。针对太阳能电池片生产过程中表面缺陷检测高精度、实时性的需求,本文提出了一种基于改进的YOLOv5s的太阳能电池片... 太阳能电池片表面缺陷的出现会严重影响太阳能转化效率,准确检测太阳能电池片表面缺陷并及时处理可以有效提高发电效率。针对太阳能电池片生产过程中表面缺陷检测高精度、实时性的需求,本文提出了一种基于改进的YOLOv5s的太阳能电池片表面缺陷检测算法。该算法首先在主干特征提取网络中用C3CA模块替换网络中C3模块,并加入CBAM注意力机制,提升网络的特征提取能力;其次,在特征融合网络中引入BiFPN网络结构,提升网络中不同语义和尺度信息的特征融合能力;最后,在输出端引入解耦头,提高了模型网络的收敛速度与检测精度。实验结果表明:改进模型在光伏电池EL数据集上平均精度均值mAP@0.5∶0.95为66.4%,相较于原网络提高了7.1%,实现了对太阳能电池片表面缺陷的快速有效定位识别,在太阳能电池工业生产过程中具有一定的实际应用价值。 展开更多
关键词 太阳能电池片 YOLOv5s 缺陷检测 注意力机制 BiFPN
在线阅读 下载PDF
改进YOLOv8的光伏电池缺陷检测算法 被引量:2
13
作者 杨丽 杨晨晨 +2 位作者 杨耿煌 段海龙 邓靖威 《电子测量技术》 北大核心 2025年第1期92-99,共8页
针对光伏电池缺陷检测在复杂背景下存在的误检、漏检等问题,提出了一种基于改进YOLOv8的光伏电池缺陷检测算法。首先,采用双向特征金字塔网络作为特征融合机制,通过自顶向下和自底向上的路径,实现多尺度特征的有效融合;其次,在颈部网络... 针对光伏电池缺陷检测在复杂背景下存在的误检、漏检等问题,提出了一种基于改进YOLOv8的光伏电池缺陷检测算法。首先,采用双向特征金字塔网络作为特征融合机制,通过自顶向下和自底向上的路径,实现多尺度特征的有效融合;其次,在颈部网络引入上下文聚合模块,使用不同空洞卷积速率的空洞卷积获取不同感受野的上下文信息,帮助模型更精准地识别微小目标,进而提升模型的目标检测性能;最后,优化边界框损失函数,并不断调试其权重因子,提高模型的收敛速度与效率。实验结果表明,与YOLOv8算法检测网络相比,本文算法的召回率和平均精确度均值分别提高了10.4%、1.8%,检测帧率达到270 fps,保证了实时检测和后续部署的轻量化要求,改进后的算法能在复杂背景下对光伏电池的缺陷进行鲁棒检测。 展开更多
关键词 光伏电池 缺陷检测 YOLOv8 注意力机制 损失函数
在线阅读 下载PDF
基于改进YOLOv8的轻量化皮革缺陷检测方法 被引量:2
14
作者 方明 张娇 +1 位作者 徐晶 王绎覃 《电子测量技术》 北大核心 2025年第1期111-118,共8页
为了解决YOLOv8参数量过大影响检测速度等问题,本文以汽车座椅皮革为样本对汽车座椅表面进行缺陷检测,提出了一种轻量化的基于YOLOv8框架的皮革缺陷检测算法。首先,将YOLOv8原本的主干网络替换成轻量化网络StarNet, StarNet通过星型运... 为了解决YOLOv8参数量过大影响检测速度等问题,本文以汽车座椅皮革为样本对汽车座椅表面进行缺陷检测,提出了一种轻量化的基于YOLOv8框架的皮革缺陷检测算法。首先,将YOLOv8原本的主干网络替换成轻量化网络StarNet, StarNet通过星型运算实现了高维和非线性特征空间的映射,从而在紧凑的网络结构和较低的能耗下展示了出色的性能和低延迟。其次,将原本的检测头替换成轻量级共享卷积检测头,通过使用共享卷积,可以大幅减少参数数量,使得模型更轻便,以便于在资源受限的设备上部署。最后,将颈部网络的C2f模块替换成C2f_Star模块,在网络更加轻量化的同时,将不同尺度的特征图进行融合,提高目标检测的准确性和鲁棒性。在自制的HSV-Leather数据集上对模型进行实验验证,结果表明,改进后的YOLOv8-Leather检测模型性能优于YOLOv8n模型。对比YOLOv8n模型,改进后的模型在参数量上降低了57%,检测速度提升了20%,模型权重降低了52%,运算量降低了53%。实验验证了改进后的模型在解决皮革表面缺陷检测问题上的可行性。 展开更多
关键词 皮革缺陷检测 YOLOv8 目标检测 轻量化 StarNet
在线阅读 下载PDF
基于重参数化的光伏电池缺陷检测算法 被引量:1
15
作者 杨丽 邓靖威 +2 位作者 段海龙 杨晨晨 李凤泉 《电子测量技术》 北大核心 2025年第5期184-192,共9页
针对光伏电池电致发光图像缺陷的复杂背景干扰不均、形状多变和缺陷多尺度等问题,提出了一种基于重参数化的光伏电池缺陷检测算法OM-Detector。首先结合广义高效层聚合网络和在线重参数化,提出了OREPANCSPELAN4模块,引入重参数化有效地... 针对光伏电池电致发光图像缺陷的复杂背景干扰不均、形状多变和缺陷多尺度等问题,提出了一种基于重参数化的光伏电池缺陷检测算法OM-Detector。首先结合广义高效层聚合网络和在线重参数化,提出了OREPANCSPELAN4模块,引入重参数化有效地通过梯度下降优化算法进行训练,在提升精度的同时降低了模型参数量,使模型轻量化;其次,在颈部网络中引入了多尺度卷积注意力模块,抑制复杂背景的干扰,提高模型检测细小缺陷的准确率;最后,结合重参数化特征提取—融合模块和多尺度卷积注意力模块,构建光伏电池缺陷检测器。使用光伏电池异常检测数据集对算法性能进行验证,实验结果表明,与YOLOv8检测网络相比,平均精度均值提升了2.5%,参数量降低了29%,推理速度加快了5.7%,优于目前的主流目标检测算法,能快速、准确地对光伏电池表面缺陷进行检测。 展开更多
关键词 缺陷检测 重参数化 注意力机制
在线阅读 下载PDF
基于改进YOLOv8n的轻量化马铃薯表面缺陷在线检测方法 被引量:5
16
作者 许英超 刘书玮 +4 位作者 王相友 吴海涛 黄杰 王恒仁 王毅 《农业工程学报》 北大核心 2025年第5期135-144,共10页
马铃薯表面缺陷是农产品分级的重要依据。为提升马铃薯表面缺陷检测精度并实现模型在移动端的快速识别,该研究提出了一种基于改进YOLOv8n的马铃薯表面缺陷检测方法DATW-YOLOv8。算法使用Dilation-wise Residual模块替换C2f中的Bottlenec... 马铃薯表面缺陷是农产品分级的重要依据。为提升马铃薯表面缺陷检测精度并实现模型在移动端的快速识别,该研究提出了一种基于改进YOLOv8n的马铃薯表面缺陷检测方法DATW-YOLOv8。算法使用Dilation-wise Residual模块替换C2f中的Bottleneck模块,并引入Dilated Reparam Block模块对C2f进行二次改进,加强细节特征提取,提高缺陷特征的提取精度;随后,引入轻量级自适应下采样(ADOWN)卷积模块,实现图像数据的有效降维,提升模型处理效率;此外,改造检测头为任务对齐动态检测头(task align dynamic detection head,TADDH),提高缺陷边界预测精度,精准聚焦缺陷关键区;最终,使用Wise-EIoU作为边界框回归损失函数,增强模型对边界模糊样本的关注度,提升缺陷边界回归精度及模型鲁棒性。试验结果表明,改进DATW-YOLOv8模型在准确率、召回率和平均精度方面分别达到95.8%、88.1%和94.3%,参数量和权重分别为1.5 M和3.6 MB。与原YOLOv8n模型相比,参数量和权重分别减少了50.0%和42.9%,同时准确率、召回率和平均精度分别提高了2.8、1.6和1.4个百分点。该方法能满足实际生产中针对缺陷马铃薯进行精准、实时检测的要求,为马铃薯表面缺陷在线检测及模型在移动端的部署提供了技术参考。 展开更多
关键词 图像识别 深度学习 无损检测 马铃薯 表面缺陷 轻量化 YOLOv8
在线阅读 下载PDF
基于改进Transformer结构的电力绝缘子运动模糊图像复原网络 被引量:1
17
作者 李鹏 常乐 +2 位作者 覃发富 孟庆伟 陈继明 《电网技术》 北大核心 2025年第6期2623-2631,I0143-I0146,共13页
针对高压输电线路巡检航拍过程中产生的电力绝缘子图像运动模糊的失真情形,影响后续绝缘子定位及缺陷检测的问题,提出了一种基于改进Transformer结构的电力绝缘子图像运动模糊复原方法。为了适应电力绝缘子航拍图像中全局与局部模糊的... 针对高压输电线路巡检航拍过程中产生的电力绝缘子图像运动模糊的失真情形,影响后续绝缘子定位及缺陷检测的问题,提出了一种基于改进Transformer结构的电力绝缘子图像运动模糊复原方法。为了适应电力绝缘子航拍图像中全局与局部模糊的复原需求,在Transformer网络结构上引入条带注意力模块,结合卷积神经网络,在减小内存空间需求和不依赖大量训练数据的同时实现高效的模糊绝缘子图像复原;同时,在网络目标函数中引入对比学习损失,充分地挖掘和利用清晰与模糊电力绝缘子图像的关联信息。构建运动模糊绝缘子图像数据集进行图像复原与缺陷检测实验,结果表明,该文的运动模糊绝缘子图像复原方法在峰值信噪比(peak signal-to-noise ratio,PSNR)和结构相似度(structure similarity index measure,SSIM)这两个指标上均高于Deblur GAN-v2、MIMO-UNet等主流算法,使用目标检测算法YOLOv5和YOLOv7对去模糊前后的绝缘子进行定位与自爆缺陷检测后显示该文方法在提升高压输电线路巡检任务中绝缘子定位与缺陷检测的准确率上具有实际应用意义。 展开更多
关键词 运动模糊图像复原 TRANSFORMER 对比学习 绝缘子及缺陷检测
在线阅读 下载PDF
基于YOLOv5s-Ghost的PCB缺陷检测系统设计 被引量:2
18
作者 冯桑 温佳旺 +1 位作者 张泳 陈兴彬 《机械设计与制造》 北大核心 2025年第3期354-357,361,共5页
这里针对常见的PCB制造缺陷,设计了一种了基于卷积神经网络的视觉检测系统。首先搭建了包括图像采集、检测结果显示与标记等模块的系统平台。通过数据采集平台拍摄PCB样本作为训练数据,使用均值滤波和数据增强技术对图像进行处理和扩充... 这里针对常见的PCB制造缺陷,设计了一种了基于卷积神经网络的视觉检测系统。首先搭建了包括图像采集、检测结果显示与标记等模块的系统平台。通过数据采集平台拍摄PCB样本作为训练数据,使用均值滤波和数据增强技术对图像进行处理和扩充,并利用SRGAN算法进行了超分辨率重建。使用GhostNET中的Ghost module对YOLOv5s进行了轻量化改进,实验结果表明,该系统具有较高的准确率,在识别准确度方面达到93%,标记准确率达到98%,同时模型的参数量压缩了48%,能够满足实际检测需求。 展开更多
关键词 机器视觉 YOLOv5 缺陷检测 轻量化
在线阅读 下载PDF
基于机器视觉的铜管锯齿伤智能识别与软件实现
19
作者 刘劲松 臧雪颖 +3 位作者 陈大勇 宋鸿武 刘昆 云红杰 《精密成形工程》 北大核心 2025年第2期130-140,共11页
目的锯齿伤作为铜管材首要的缺陷,会造成管材后续二次加工开裂和服役破裂等严重问题,对缺陷进行高效、定量化的统计及分析是产品质量管控的关键性难点,也是铜管加工行业的空白。针对这一问题,基于机器视觉方法,设计并开发了铜管材锯齿... 目的锯齿伤作为铜管材首要的缺陷,会造成管材后续二次加工开裂和服役破裂等严重问题,对缺陷进行高效、定量化的统计及分析是产品质量管控的关键性难点,也是铜管加工行业的空白。针对这一问题,基于机器视觉方法,设计并开发了铜管材锯齿伤缺陷智能识别算法和系统。方法首先,基于自适应阈值算法,设计采用一系列的降噪算法,实现锯齿伤管材横截面轮廓识别。其次,定义了锯齿伤缺陷占比的计算规则,通过算法优化,实现了开口和闭口2种锯齿伤的成功识别和缺陷占比的计算。最后,借助Python语言构建了图形用户界面(GUI)系统,完成了锯齿伤缺陷统计的软件化。结果通过对智能识别算法和实验获得的锯齿伤缺陷占比进行对比,发现相对误差保持在1%~2%,证明缺陷智能识别算法具有可行性和高精度。结论该系统可以有效地实现铜管锯齿伤缺陷识别和占比的定量、高效统计与分析。 展开更多
关键词 精密铜管 机器视觉 锯齿伤 缺陷识别 缺陷占比
在线阅读 下载PDF
基于扩散模型检测的高铁接触网绝缘子缺陷语义描述方法 被引量:1
20
作者 陈永 安卓奥博 周建宇 《电工技术学报》 北大核心 2025年第13期4100-4111,共12页
高铁接触网绝缘子作为高速铁路牵引供电的重要装置,可为接触网提供电气部件绝缘和腕臂结构支撑,其安全性对于高速铁路行车至关重要。针对绝缘子检测时易受复杂环境背景干扰,导致缺陷检测精度低以及无法提供缺陷语义描述的问题,该文提出... 高铁接触网绝缘子作为高速铁路牵引供电的重要装置,可为接触网提供电气部件绝缘和腕臂结构支撑,其安全性对于高速铁路行车至关重要。针对绝缘子检测时易受复杂环境背景干扰,导致缺陷检测精度低以及无法提供缺陷语义描述的问题,该文提出一种基于扩散模型检测的绝缘子缺陷描述方法。首先,构建大核空间选择特征提取网络,加强绝缘子缺陷特征信息的提取能力;其次,基于扩散模型设计融合扩散机制的检测解码器,并对解码器生成的噪声框进行逆向贝叶斯扩散,还原绝缘子真值框的预测,提高模型的抗背景干扰能力;最后,设计交叉注意力机制的编码器和解码器,实现图像与文本的跨模态映射,并通过文本过滤机制驱动的多模态语言视觉预训练(BLIP)模型,完成绝缘子缺陷文本描述输出。实验结果表明,所提绝缘子缺陷检测模型的平均准确度达到93.04%,相较于DTER和Faster RCNN的mAP0.5分别提升4.63%和5.78%,且F1-score高达82.91%,平均双语评估替换评价指标(BLEU)和基于精确率的图像描述评价指标(CIDEr)分别达到83.51%和1.94。与其他方法相比,具有更高的检测精度和缺陷语义描述准确性,能够满足对高速铁路绝缘子缺陷的检测需求。 展开更多
关键词 高铁接触网 绝缘子缺陷检测 缺陷语义描述 扩散模型 交叉注意力机制
在线阅读 下载PDF
上一页 1 2 250 下一页 到第
使用帮助 返回顶部