期刊文献+
共找到56,910篇文章
< 1 2 250 >
每页显示 20 50 100
基于音视频信息融合与Self-Attention-DSC-CNN6网络的鲈鱼摄食强度分类方法 被引量:1
1
作者 李道亮 李万超 杜壮壮 《农业机械学报》 北大核心 2025年第1期16-24,共9页
摄食强度识别分类是实现水产养殖精准投喂的重要环节。现有的投喂方式存在过度依赖人工经验判断、投喂量不精确、饲料浪费严重等问题。基于多模态融合的鱼类摄食程度分类能够综合不同类型的数据(如:视频、声音和水质参数),为鱼群的投喂... 摄食强度识别分类是实现水产养殖精准投喂的重要环节。现有的投喂方式存在过度依赖人工经验判断、投喂量不精确、饲料浪费严重等问题。基于多模态融合的鱼类摄食程度分类能够综合不同类型的数据(如:视频、声音和水质参数),为鱼群的投喂提供更加全面精准的决策依据。因此,提出了一种融合视频和音频数据的多模态融合框架,旨在提升鲈鱼摄食强度分类性能。将预处理后的Mel频谱图(Mel Spectrogram)和视频帧图像分别输入到Self-Attention-DSC-CNN6(Self-attention-depthwise separable convolution-CNN6)优化模型进行高层次的特征提取,并将提取的特征进一步拼接融合,最后将拼接后的特征经分类器分类。针对Self-Attention-DSC-CNN6优化模型,基于CNN6算法进行了改进,将传统卷积层替换为深度可分离卷积(Depthwise separable convolution,DSC)来达到减少计算复杂度的效果,并引入Self-Attention注意力机制以增强特征提取能力。实验结果显示,本文所提出的多模态融合框架鲈鱼摄食强度分类准确率达到90.24%,模型可以有效利用不同数据源信息,提升了对复杂环境中鱼群行为的理解,增强了模型决策能力,确保了投喂策略的及时性与准确性,从而有效减少了饲料浪费。 展开更多
关键词 鲈鱼 摄食强度分类 多模态融合 self-Attention-DSC-CNN6
在线阅读 下载PDF
基于TV正则化约束的Self2Self地震数据插值去噪一体化方法
2
作者 张蕴 杨锴 王本锋 《地球物理学报》 北大核心 2025年第9期3575-3587,共13页
实际采集的地震数据不可避免地受到随机噪声的干扰,并常常伴随着数据缺失,严重降低了地震数据的信噪比与横向连续性,继而降低后续地震数据处理及反演的精度.本文基于Self2Self无监督学习框架,针对含噪非规则地震数据,设计无监督地震数... 实际采集的地震数据不可避免地受到随机噪声的干扰,并常常伴随着数据缺失,严重降低了地震数据的信噪比与横向连续性,继而降低后续地震数据处理及反演的精度.本文基于Self2Self无监督学习框架,针对含噪非规则地震数据,设计无监督地震数据插值去噪一体化方法,并基于地震道之间的相关性优化数据采样策略,对含噪的非规则地震数据进行整道伯努利采样,构建训练数据集;为降低随机噪声对插值重建的负面影响,在损失函数中引入全变分正则化项,确保恢复地震信号具有良好的横向连续性.针对异常噪声干扰,探讨了异常噪声识别-剔除策略,结合插值去噪一体化方法可以有效提高地震数据质量.不同数值算例验证了无监督Self2Self方法在地震数据插值重建及噪声衰减中的有效性,为后续地震数据处理和解释提供良好的数据支撑. 展开更多
关键词 随机噪声衰减 数据重建 无监督学习 self2self方法 TV正则化
在线阅读 下载PDF
基于Self-Attention和TextCNN-BiLSTM的中文评论文本情感分析模型 被引量:3
3
作者 龙宇 李秋生 《石河子大学学报(自然科学版)》 北大核心 2025年第1期111-121,共11页
目前关于中文评论文本的情感分类方法大都无法充分捕捉到句子的全局语义信息,同时也在长距离的语义连接或者情感转折理解上具有局限性,因而导致情感分析的准确度不高。针对这个问题,本文提出一种融合SelfAttention和TextCNN-BiLSTM的文... 目前关于中文评论文本的情感分类方法大都无法充分捕捉到句子的全局语义信息,同时也在长距离的语义连接或者情感转折理解上具有局限性,因而导致情感分析的准确度不高。针对这个问题,本文提出一种融合SelfAttention和TextCNN-BiLSTM的文本情感分析方法。该方法首先采用文本卷积神经网络(TextCNN)来提取局部特征,并利用双向长短期记忆网络(BiLSTM)来捕捉序列信息,从而综合考虑了全局和局部信息,在特征融合阶段,再采用自注意力机制来动态地融合不同层次的特征表示,对不同尺度特征进行加权,从而提高重要特征的响应。实验结果表明,所提出的模型在家电商品中文评论语料和谭松波酒店评论语料数据集上的准确率分别达到93.79%和90.05%,相较于基准模型分别提高0.69%~3.59%和4.44%~11.70%,优于传统的基于卷积神经网络(Convolutional Neural Networks, CNN)、BiLSTM或CNN-BiLSTM等的情感分析模型。 展开更多
关键词 自注意力机制 中文评论文本 深度学习 情感分析
在线阅读 下载PDF
基于VMD-Self-attention-LSTM的水闸深基坑变形智能预测方法 被引量:2
4
作者 张伟 仇建春 +5 位作者 夏国春 姚兆仁 吴昊 刘占午 王昱锦 朱新宇 《水电能源科学》 北大核心 2025年第1期99-102,196,共5页
针对水闸深基坑变形监测数据具有非稳定性的特点,提出了基于VMD-Self-attention-LSTM的水闸深基坑变形预测方法。该方法分为三个主要模块,第一模块采用VMD算法自适应调整变形分解模数,将原始变形数据分解为若干具有明显周期规律的分量,... 针对水闸深基坑变形监测数据具有非稳定性的特点,提出了基于VMD-Self-attention-LSTM的水闸深基坑变形预测方法。该方法分为三个主要模块,第一模块采用VMD算法自适应调整变形分解模数,将原始变形数据分解为若干具有明显周期规律的分量,有效解决变形数据的非稳定性,为提升变形预测精度奠定初步基础;第二模块,在传统LSTM算法基础上进一步发展Self-attention-LSTM方法,提升模型对基坑变形样本的时序关系挖掘能力,进而提升基坑变形预测精度;第三模块,将各分量对应的变形预测结果重构得到最终预测值。实例分析可知,所提方法有效解决了非稳定特性给变形预测精度带来的不利影响,与VMD-LSTM、Self-attention-LSTM、LSTM等深度学习方法相比,VMD-Self-attention-LSTM的预测精度最大提升41.49%,与BP、ELM等传统机器学习算法相比,预测精度最大提升50.43%,为水闸深基坑安全监控模型的构建提供了新思路。 展开更多
关键词 水闸深基坑 变形预测 VMD self-attention-LSTM
在线阅读 下载PDF
Microstructure and pitting corrosion of shielded metal arc welded high nitrogen stainless steel 被引量:8
5
作者 RAFFI MOHAMMED G.MADHUSUDHAN REDDY K.SRINIVASA RAO 《Defence Technology(防务技术)》 SCIE EI CAS CSCD 2015年第3期237-243,共7页
The present work is aimed at studying the microstructure and pitting corrosion behaviour of shielded metal arc welded high nitrogen steel made of Cromang-N electrode. Basis for selecting this electrode is to increase ... The present work is aimed at studying the microstructure and pitting corrosion behaviour of shielded metal arc welded high nitrogen steel made of Cromang-N electrode. Basis for selecting this electrode is to increase the solubility of nitrogen in weld metal due to high chromium and manganese content. Microscopic studies were carried out using optical microscopy(OM) and field emission scanning electron microscopy(FESEM). Energy back scattered diffraction(EBSD) method was used to determine the phase analysis, grain size and orientation image mapping. Potentio-dynamic polarization testing was carried out to study the pitting corrosion resistance in aerated 3.5% NaCl environment using a GillAC electrochemical system. The investigation results showed that the selected Cr-Mn-N type electrode resulted in a maximum reduction in delta-ferrite and improvement in pitting corrosion resistance of the weld zone was attributed to the coarse austenite grains owing to the reduction in active sites of the austenite/delta ferrite interface and the decrease in galvanic interaction between austenite and delta-ferrite. 展开更多
关键词 耐点蚀性能 高氮不锈钢 焊缝金属 电弧焊接 研究组织 屏蔽 场发射扫描电子显微镜 奥氏体晶粒
在线阅读 下载PDF
Low-value biomass-derived carbon composites for electromagnetic wave absorption and shielding: A review
6
作者 Sumanta Sahoo Rajesh Kumar Sung Soo Han 《新型炭材料(中英文)》 北大核心 2025年第2期293-316,共24页
The rising concern over electromagnetic (EM) pollution is re-sponsible for the rapid progress in EM interference (EMI) shielding and EM wave absorption in the last few years, and carbon materials with a large sur-face... The rising concern over electromagnetic (EM) pollution is re-sponsible for the rapid progress in EM interference (EMI) shielding and EM wave absorption in the last few years, and carbon materials with a large sur-face area and high porosity have been investigated. Compared to other car-bon materials, biomass-derived carbon (BC) are considered efficient and eco-friendly materials for this purpose. We summarize the recent advances in BC materials for both EMI shielding and EM wave absorption. After a brief overview of the synthesis strategies of BC materials and a precise out-line of EM wave interference, strategies for improving their EMI shielding and EM wave absorption are discussed. Finally, the existing challenges and the future prospects for such materials are briefly summarized. 展开更多
关键词 Biomass carbon COMPOSITES Dielectric loss EMI shielding EM wave absorption
在线阅读 下载PDF
Design and implementation of a high-performance miniaturized multi-layer magnetic shielding spherical shell with minimal pores based on target magnetic field
7
作者 Wanqing Li Wei Li +2 位作者 Shaotao Qu Shuyi Zhang Jingyang Xu 《Defence Technology(防务技术)》 2025年第3期99-108,共10页
Achieving the spin-exchange relaxation-free(SERF)state in atomic comagnetometers(ACMs)necessitates a stable and weak magnetic environment.This paper presents the design of a miniaturized permalloy magnetic shielding s... Achieving the spin-exchange relaxation-free(SERF)state in atomic comagnetometers(ACMs)necessitates a stable and weak magnetic environment.This paper presents the design of a miniaturized permalloy magnetic shielding spherical shell(MSSS)with minimal apertures,tailored to meet these requirements.By employing a combination of analytical solutions and finite element analysis(FEA),we achieved superior magnetic shielding while maintaining a compact form factor.The analytical solution for the shielding factor indicated that a four-layer permalloy sphere shell with optimized air gaps was necessary.A numerical analysis model of the MSSS was developed and validated using COMSOL software,confirming the suitability of the air gaps.The size,shape,and orientation of the openings in the perforated sphere shell were meticulously designed and optimized to minimize residual magnetism.The optimal structure was fabricated,resulting in triaxial shielding factors of 47619,52631,and 21739,meeting the anticipated requirements.A comparison of simulation results with experimental tests demonstrated the efficacy of the design methodology.This study has significant implications for ultrasensitive magnetic field detection devices requiring weak magnetic field environments,such as atomic gyroscopes,magnetometers,atomic interferometers,and atomic clocks. 展开更多
关键词 Magnetic shielding Spherical shield Finite element analysis(FEA) shielding effectiveness Spin-exchange relaxation-free(SERF) comagnetometer
在线阅读 下载PDF
Bismuth's role in quaternary thulium doped tellurite glasses:Achieving optical clarity and radiation shielding
8
作者 N.A.M.Rusni N.A.Abdul-Manaf +3 位作者 N.Ahmad L.Hasnimulyati W.Y.W.Yusoff A.Azuraida 《Defence Technology(防务技术)》 2025年第9期134-144,共11页
This study explores the impact of bismuth oxide(Bi_(2)O_(3))on the optical and radiation shielding properties of transparent,lead-free thulium-doped bismuth borotellurite radiation shielding glass.The investigated gla... This study explores the impact of bismuth oxide(Bi_(2)O_(3))on the optical and radiation shielding properties of transparent,lead-free thulium-doped bismuth borotellurite radiation shielding glass.The investigated glass composition follows the formula[(TeO_(2))_(75)(B_(2)O_(3))_(25)]_(98-x)(Bi_(2)O_(3))_x[Tm_(2)O_(3)]_(2),where x=0 mol%,5 mol%,10 mol%,15 mol%,20 mol%,25 mol%,and 30 mol%.All glass samples remain transparent,with an optical bandgap(E_(opt))exceeding 3.1 e V,ensuring visible light transmission.Radiation shielding data from Phy-X and XCom reveal interactions of the photoelectric effect,Compton scattering,and pair production,with minimal relative difference in mass attenuation coefficient(MAC)which is between0.05 and 0.56.At 0.662 Me V photon energy,the 20 mol%and 25 mol%Bi_(2)O_(3)glasses exhibit significantly higher Phy-X MAC values than other samples,except RS 520 glass,which contains 71%Pb O.Despite incorporating only up to 25 mol%Bi_(2)O_(3),these glasses outperform others in density,half-value layer(HVL),and mean free path(MFP).Correlating E_(opt)and MAC,the 20 mol%Bi_(2)O_(3)glass is the best candidate for transparent radiation shielding glass due to its wide optical bandgap which prevents ionization of trapped holes.Significantly,the linkage between MFP and molar refraction was also discovered based on the particle size influence on both parameters. 展开更多
关键词 Thulium-doped bismuth borotellurite glasses OPTICAL Radiation shielding LEAD-FREE Phy-X XCom
在线阅读 下载PDF
Longitudinal structural vulnerability analysis of shield tunnels under adjacent excavation disturbances
9
作者 PENG Zhu SHI Cheng-hua +2 位作者 WANG Zu-xian LEI Ming-feng PENG Li-min 《Journal of Central South University》 2025年第6期2256-2272,共17页
This paper proposes a longitudinal vulnerability-based analysis method to evaluate the impact of foundation pit excavation on shield tunnels,accounting for geological uncertainties.First,the shield tunnel is modeled a... This paper proposes a longitudinal vulnerability-based analysis method to evaluate the impact of foundation pit excavation on shield tunnels,accounting for geological uncertainties.First,the shield tunnel is modeled as an Euler Bernoulli beam resting on the Pasternak foundation incorporating variability in subgrade parameters along the tunnel’s length.A random analysis method using random field theory is introduced to evaluate the tunnel’s longitudinal responses to excavation.Next,a risk assessment index system is established.The normalized relative depth between the excavation and the shield tunnel is used as a risk index,while the maximum longitudinal deformation,the maximum circumferential opening,and the maximum longitudinal bending moment serve as performance indicators.Based on these,a method for analyzing the longitudinal fragility of shield tunnels under excavation-induced disturbances is proposed.Finally,the technique is applied to a case study involving a foundation pit excavation above a shield tunnel,which is the primary application scenario of this method.Vulnerability curves for different performance indicators are derived,and the effects of tunnel stiffness and subgrade stiffness on the tunnel vulnerability are explored.The results reveal significant differences in vulnerability curves depending on the performance index used.Compared to the maximum circumferential opening and the maximum longitudinal bending moment,selecting the maximum longitudinal deformation as the control index better ensures the tunnel’s usability and safety under excavation disturbances.The longitudinal vulnerability of the shield tunnel nonlinearly decreases with the increase of the tunnel stiffness and subgrade stiffness,and the subgrade stiffness has a more pronounced effect.Parametric analyses suggest that actively reinforcing the substratum is more effective on reducing the risk of tunnel failure due to adjacent excavations than passive reinforcement of the tunnel structure. 展开更多
关键词 shield tunnel foundation pit excavation foundation variability random field VULNERABILITY
在线阅读 下载PDF
Review of bumper materials for spacecraft shield against orbital debris hypervelocity impact
10
作者 Siyuan Ren Pinliang Zhang +6 位作者 Qiang Wu Qingming Zhang Zizheng Gong Guangming Song Renrong Long Liangfei Gong Mingze Wu 《Defence Technology(防务技术)》 2025年第3期137-177,共41页
It is widely known that the hypervelocity impact of orbital debris can cause serious damage to spacecraft,and enhancing the impact resistance is the great concern of spacecraft shield design.This paper provides a comp... It is widely known that the hypervelocity impact of orbital debris can cause serious damage to spacecraft,and enhancing the impact resistance is the great concern of spacecraft shield design.This paper provides a comprehensive overview of advances in the development of bumper materials for spacecraft shield applications.In particular,the protective mechanism and process of the bumper using different materials against hypervelocity impact are reviewed and discussed.The advantages and disadvantages of each material used in shield were discussed,and the performance under hypervelocity impact was given according to the specific configuration.This review provides the useful reference and basis for researchers and engineers to create bumper materials for spacecraft shield applications,and the contemporary challenges and future directions for bumper materials for spacecraft shield were presented. 展开更多
关键词 Orbital debris Spacecraft shield Hypervelocity impact Bumper materials Protective mechanism
在线阅读 下载PDF
Performance analysis of fluid-core Whipple shields under hypervelocity impact at different projectile speeds
11
作者 Anand Pai Marcos Rodriguez-Millan +3 位作者 Selim Gürgen Adithya Piccholiya Nishant Mujumdar Satish Shenoy B 《Defence Technology(防务技术)》 2025年第5期32-45,共14页
Whipple shields as sacrificial bumpers,safeguard the satellites against extremely fast,different-sized projectiles traveling through space in the low earth orbit.Typical Whipple shields comprise a front and rear plate... Whipple shields as sacrificial bumpers,safeguard the satellites against extremely fast,different-sized projectiles traveling through space in the low earth orbit.Typical Whipple shields comprise a front and rear plate,separated by a gap or space.Recent advancements have explored the use of foam,cellular cores,and alternative materials such as ceramics instead of aluminium for the plates.In the current work,the effect of including fluid cores(air/water)sandwiched between the front and rear plates,on the response to hypervelocity impact was explored through a numerical approach.The numerical simulation consisted of hypervelocity impact by a 2 mm diameter,stainless steel projectile,launched at speeds of 3 e9 km/s with a normal impact trajectory towards the Whipple shield.The front and rear bumpers,made of AA6061-T6,were each 1 mm thick.A space of 10 mm was taken between the plates(occupied by fluid).The key metrics analyzed were the perforation characteristics,stages of the debris cloud generation and propagation,energy variations(internal,kinetic and plastic work),temperature variations,and the fragmentation summary.From the computational analysis,employing water-core in Whipple shields could prevent the rear bumper perforation till 6 km/s,lower the peak temperatures at the front bumper perforation zones and debris tip,and generate fewer,larger fragments. 展开更多
关键词 Whipple shields Fluid-filled core Hyper velocity impact Debris cloud Perforation characteristics Numerical simulation
在线阅读 下载PDF
Damage prediction of rear plate in Whipple shields based on machine learning method
12
作者 Chenyang Wu Xiangbiao Liao +1 位作者 Lvtan Chen Xiaowei Chen 《Defence Technology(防务技术)》 2025年第8期52-68,共17页
A typical Whipple shield consists of double-layered plates with a certain gap.The space debris impacts the outer plate and is broken into a debris cloud(shattered,molten,vaporized)with dispersed energy and momentum,wh... A typical Whipple shield consists of double-layered plates with a certain gap.The space debris impacts the outer plate and is broken into a debris cloud(shattered,molten,vaporized)with dispersed energy and momentum,which reduces the risk of penetrating the bulkhead.In the realm of hypervelocity impact,strain rate(>10^(5)s^(-1))effects are negligible,and fluid dynamics is employed to describe the impact process.Efficient numerical tools for precisely predicting the damage degree can greatly accelerate the design and optimization of advanced protective structures.Current hypervelocity impact research primarily focuses on the interaction between projectile and front plate and the movement of debris cloud.However,the damage mechanism of debris cloud impacts on rear plates-the critical threat component-remains underexplored owing to complex multi-physics processes and prohibitive computational costs.Existing approaches,ranging from semi-empirical equations to a machine learningbased ballistic limit prediction method,are constrained to binary penetration classification.Alternatively,the uneven data from experiments and simulations caused these methods to be ineffective when the projectile has irregular shapes and complicate flight attitude.Therefore,it is urgent to develop a new damage prediction method for predicting the rear plate damage,which can help to gain a deeper understanding of the damage mechanism.In this study,a machine learning(ML)method is developed to predict the damage distribution in the rear plate.Based on the unit velocity space,the discretized information of debris cloud and rear plate damage from rare simulation cases is used as input data for training the ML models,while the generalization ability for damage distribution prediction is tested by other simulation cases with different attack angles.The results demonstrate that the training and prediction accuracies using the Random Forest(RF)algorithm significantly surpass those using Artificial Neural Networks(ANNs)and Support Vector Machine(SVM).The RF-based model effectively identifies damage features in sparsely distributed debris cloud and cumulative effect.This study establishes an expandable new dataset that accommodates additional parameters to improve the prediction accuracy.Results demonstrate the model's ability to overcome data imbalance limitations through debris cloud features,enabling rapid and accurate rear plate damage prediction across wider scenarios with minimal data requirements. 展开更多
关键词 Damage prediction of rear plate Cumulative effect of debris cloud Whipple shield Machine learning Random forest
在线阅读 下载PDF
Self-Guide:一种基于自我规划的大语言模型推理增强方法
13
作者 刘艺彬 刘正皓 +6 位作者 闫宇坤 于是 王硕 杨麟儿 陈慧敏 谷峪 于戈 《中文信息学报》 北大核心 2025年第4期55-66,共12页
尽管大语言模型在自然语言处理任务中取得显著进展,但其在复杂问题推理等领域还面临着认知负荷问题,即大语言模型在推理过程中需要记忆并处理大量信息。因此,如何有效地减少大语言模型推理过程中的认知负荷,缓解推理过程中可能出现的认... 尽管大语言模型在自然语言处理任务中取得显著进展,但其在复杂问题推理等领域还面临着认知负荷问题,即大语言模型在推理过程中需要记忆并处理大量信息。因此,如何有效地减少大语言模型推理过程中的认知负荷,缓解推理过程中可能出现的认知过载,是一个亟待解决的问题。对此该文提出了Self-Guide方法,用于增强语言模型的推理能力。该方法通过指引大语言模型生成常识知识和推理指导,让大语言模型基于自我规划来增强其推理能力,并通过与推理链结合的方式对模型的推理过程进行校准。与现有方法不同的是,该文在不对大语言模型进行微调或使用外部工具的情况下,显著提升了大语言模型的推理性能。实验结果表明,Self-Guide方法在四种常见推理任务上性能显著优于基线方法,同时相比传统的推理链模型,Self-Guide方法在推理能力较弱的模型上也具有良好的泛化性能。通过结合大语言模型的自我规划和推理能力,Self-Guide方法为提升语言模型的推理能力提供了一种新的有效途径。 展开更多
关键词 大语言模型 自我引导 推理增强 认知负荷
在线阅读 下载PDF
一种基于用户行为的Self集构造和演化方法 被引量:11
14
作者 梁意文 李俊涛 +1 位作者 郭学理 康立山 《计算机应用研究》 CSCD 北大核心 2001年第9期7-9,14,共4页
对计算机安全免疫系统中Self集构造的可能性和特征码选取的原则进行了探讨性的分析 ,提出了一种基于用户行为的Self集构造方法。Self集中的特征码包含了用户的操作、操作对象和操作参数。
关键词 演化计算 入侵检测 self 计算机安全免疫系统 计算机网络 防火墙
在线阅读 下载PDF
层流区Shields曲线分布规律试验
15
作者 洪安宇 王正中 赵延风 《水科学进展》 EI CAS CSCD 北大核心 2015年第6期837-845,共9页
为探明层流条件下Shields曲线的分布规律,运用电荷耦合组件(Charge Coupled Device,CCD)成像技术结合激光扫描及计算机图像处理技术,在水和甘油混合液中对无黏性均匀玻璃颗粒的起动拖曳力进行测量,并根据层流绕流理论建立了无黏性均匀... 为探明层流条件下Shields曲线的分布规律,运用电荷耦合组件(Charge Coupled Device,CCD)成像技术结合激光扫描及计算机图像处理技术,在水和甘油混合液中对无黏性均匀玻璃颗粒的起动拖曳力进行测量,并根据层流绕流理论建立了无黏性均匀颗粒的滚动起动模型,提出层流条件下无黏性均匀颗粒的量纲一起动拖曳力公式。试验观测发现在层流范围内希尔兹数呈规则的带状分布,流体作用引起颗粒床表面粗化并导致床面颗粒突起减小,使颗粒起动拖曳力增大1倍以上。结果表明颗粒床表面结构性状直接影响着颗粒起动拖曳力的大小,在层流区Shields曲线具有带状分布特性。 展开更多
关键词 shields曲线 层流 CCD成像技术 无黏性均匀颗粒 颗粒床表面结构
在线阅读 下载PDF
基于CEEMDAN和HBA-BiGRU-SelfAttention的短期负荷预测
16
作者 朱婷 颜七笙 《南京信息工程大学学报》 北大核心 2025年第4期478-493,共16页
针对电力负荷数据存在非线性、时序性等多方面因素导致的预测精度不足等问题,本文提出一种基于CEEMDAN和HBA-BiGRU-SelfAttention的短期负荷预测模型.首先,采用随机森林(RF)算法对气象因素进行特征提取,在保证数据特征的同时,降低数据... 针对电力负荷数据存在非线性、时序性等多方面因素导致的预测精度不足等问题,本文提出一种基于CEEMDAN和HBA-BiGRU-SelfAttention的短期负荷预测模型.首先,采用随机森林(RF)算法对气象因素进行特征提取,在保证数据特征的同时,降低数据的复杂度;其次,采用自适应噪声完备集合经验模态分解(CEEMDAN)算法对原始负荷数据进行分解,得到若干较为平稳的模态分量;然后,将经过特征提取的气象因素和模态分量作为输入数据,利用BiGRU(双向门控循环单元)-SelfAttention(自注意力机制)模型进行预测,并针对BiGRU-SelfAttention模型的超参数难以选取最优解的问题,引入蜜獾算法(HBA)对BiGRU-SelfAttention模型的超参数进行寻优;最后,将子序列预测结果叠加,得到最终预测结果.以某地实际电力负荷数据为数据集进行对比试验,结果表明,本文所提出的模型具有较高的预测精度,可以为电力系统稳定运行提供可靠依据. 展开更多
关键词 短期负荷预测 随机森林 自适应噪声完备集合经验模态分解 蜜獾算法 双向门控循环单元 自注意力机制
在线阅读 下载PDF
基于互联网和self-training的中文问答模式学习 被引量:2
17
作者 李志圣 孙越恒 +1 位作者 何丕廉 候越先 《计算机应用》 CSCD 北大核心 2008年第6期1575-1577,1581,共4页
在已有的问答模式学习中,模式定义和候选答案评分偏于简单,而且学习过程依赖于人工标定语料。通过挖掘W eb文本中动、名词序列的骨架模式,用以扩充模式定义;将self-train ing学习机制引入问答模式学习:用一对训练语料进行初始学习,通过... 在已有的问答模式学习中,模式定义和候选答案评分偏于简单,而且学习过程依赖于人工标定语料。通过挖掘W eb文本中动、名词序列的骨架模式,用以扩充模式定义;将self-train ing学习机制引入问答模式学习:用一对训练语料进行初始学习,通过互联网搜索,自动选择可靠程度较高的问答对,重新训练;扩充了启发规则,改进候选答案的评分方法。实验结果表明:所提出的问答模式学习方法能有效地提高中文问答系统的性能。 展开更多
关键词 互联网 问答模式 self-TRAINING 机器学习
在线阅读 下载PDF
开放存取自存档(self-archiving)的版权问题分析 被引量:12
18
作者 秦珂 《图书与情报》 CSSCI 2008年第1期103-114,105,共4页
无论是自存档预印本,还是后印本都受到版权问题的制约。目前,国际期刊界对自存档的版权政策有所放宽。为了更有利于开放存取事业的发展,在建构自存档版权制度中应尊重作者的自存档权利、满足期刊社的合理要求、完善自存档实现模式、开... 无论是自存档预印本,还是后印本都受到版权问题的制约。目前,国际期刊界对自存档的版权政策有所放宽。为了更有利于开放存取事业的发展,在建构自存档版权制度中应尊重作者的自存档权利、满足期刊社的合理要求、完善自存档实现模式、开发自存档政策指引工具、协调期刊社和自存档系统之间的版权政策。 展开更多
关键词 开放存取 自存档 预印本 后印本 版权
在线阅读 下载PDF
融合self-attention机制的卷积神经网络文本分类模型 被引量:22
19
作者 邵清 马慧萍 《小型微型计算机系统》 CSCD 北大核心 2019年第6期1137-1141,共5页
传统的文本分类算法采用词向量表示文本,忽视了上下文语境中词义的变化.本文通过引入self-attention机制处理词向量,提出一种卷积神经网络模型与关键词提取技术相结合的文本分类模型.该模型对文档进行self-attention操作,以抽取关键信息... 传统的文本分类算法采用词向量表示文本,忽视了上下文语境中词义的变化.本文通过引入self-attention机制处理词向量,提出一种卷积神经网络模型与关键词提取技术相结合的文本分类模型.该模型对文档进行self-attention操作,以抽取关键信息,构建文档特征图,根据卷积神经网络模型和关键词提取技术实现特征向量的分类.在真实数据集上进行性能分析,并与循环神经网络模型、长短时记忆网络模型进行比较,结果表明该分类模型有效地提高了分类的准确性. 展开更多
关键词 文本分类 卷积神经网络 自注意力机制 关键词提取技术
在线阅读 下载PDF
从“self-discipline”和“self-regulation”的辨析看自律的内涵
20
作者 刘刚 《中国注册会计师》 2004年第4期79-79,共1页
关键词 self—discipline” self—regulation” 自律 英语 注册会计师行业 行业协会
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