检索结果-南通市图书馆

Acta Metallurgica Sinica(English Letters) 2018年第12期31卷 1258-1264页

作者： Xiao-Pei Li song-zhu kure-chu Toru Ogasawara Hitoshi Yashiro Hai-Bo Wang Zi-Zhen Xu Xiao-Hui Li Guo-Lin song Guo-Yi Tang Advanced Materials Institute Graduate School at ShenzhenTsinghua University School of Materials Science and Engineering Tsinghua University Department of Chemistry and Bioengineering Iwate University

By using a novel surface modification technique named ultrasonic-electropulsing coupling rolling （UECR） process on an Al-Si casting alloy rod, the surface of material was smoothened significantly. Meanwhile, a strengthened layer with agradient change in hardness was obtained in the outer surface, corresponding to a homogeneous gradient nano-/micro-structure. The thickness of nanometer-thick laminated structures was at least 40 μm, which was much thicker thanconventional ultrasonic rolling process. During UECR, the formation of the well-defined nanocrystalline structure wasattributed to the high strain rate and simultaneous annealing process realized by ultrasonic impact and electropulsingtreatment.

关键词： Nanocrystalline materials Indentation and hardness Al-Si alloy Microstructure

Domain generalization with semi-supervised learning for people-centric activity recognition

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2025年第1期68卷 171-188页

作者： Jing LIU Wei zhu Di LI Xing HU Liang song Academy for Engineering & Technology Fudan University Shanghai East-bund Research Institute on Networking Systems of AI School of Optoelectronic Information and Computer Engineering University of Shanghai for Science & Technology

People-centric activity recognition is one of the most critical technologies in a wide range of real-world applications,including intelligent transportation systems, healthcare services, and brain-computer interfaces. Large-scale data collection and annotation make the application of machine learning algorithms prohibitively expensive when adapting to new tasks. One way of circumventing this limitation is to train the model in a semi-supervised learning manner that utilizes a percentage of unlabeled data to reduce the labeling burden in prediction tasks. Despite their appeal, these models often assume that labeled and unlabeled data come from similar distributions, which leads to the domain shift problem caused by the presence of distribution gaps. To address these limitations, we propose herein a novel method for people-centric activity recognition,called domain generalization with semi-supervised learning(DGSSL), that effectively enhances the representation learning and domain alignment capabilities of a model. We first design a new autoregressive discriminator for adversarial training between unlabeled and labeled source domains, extracting domain-specific features to reduce the distribution gaps. Second, we introduce two reconstruction tasks to capture the task-specific features to avoid losing information related to representation learning while maintaining task-specific consistency. Finally, benefiting from the collaborative optimization of these two tasks, the model can accurately predict both the domain and category labels of the source domains for the classification task. We conduct extensive experiments on three real-world sensing datasets. The experimental results show that DGSSL surpasses the three state-of-the-art methods with better performance and generalization.

关键词： activity recognition deep learning domain generalization semi-supervised learning adversarial training

同方期刊数据库

Effect of order–disorder transition on thermodynamic and electronic properties of σ and χ phases in W–Re alloy: the first-principles calculation

在线全文

学校读者我要写书评

暂无评论

Tungsten 2025年第1期7卷 195-213页

作者： Zhi-Peng zhu Dian Jia William Yi Wang Jun-Lei Yin Xing-Yu Gao Shu-Feng Yang Hai-Feng song Jin-Shan Li State Key Laboratory of Solidification Processing Northwestern Polytechnical University Laboratory of Computational Physics Institute of Applied Physics and Computational Mathematics School of Metallurgical and Ecological Engineering University of Science and Technology Beijing

The Tungsten-Rhenium(W-Re) alloys,celebrated for their high melting point,strength at elevated temperatures,electrical resistivity,and radiation resistance,have been widely utilized in high-temperature components,aerospace,electronics,and nuclear *** constituents of the topologically close-packed(TCP) phases,the sigma phase(σ) and chi phase(χ) formed within W-Re alloys wield considerable influence on the mechanical properties and the stability of the *** calculations were utilized in the present work to explore the structural,thermodynamic,and electronic properties of both ordered and disordered configurations within the σ and χ phases,culminating in a systematic elucidation of the higher phase stability exhibited by the ordered *** is found that the bulk modulus of these two phases is directly proportional to the concentration of Re in the alloy,while the equilibrium volume is inversely *** thermodynamic parameters of the σ and χ phases are calculated via the mean-field potential *** similar trends observed in the isobaric heat capacity,enthalpy increment,and entropy change curves for these two phases suggest they possess comparable thermodynamic *** is noteworthy that the contribution of ionic vibrations predominantly affects the isobaric heat capacity,while the contribution of thermal electronic excitations increases linearly with *** the structure and thermodynamic properties of TCP phases in W-Re alloys at low temperatures has profound significance for optimizing material performance,microstructures features,establishing theoretical foundations,and predicting material behavior.

关键词： First-principles calculations Phase stability TCP phases Thermodynamic properties W–Re alloy

Wetland and habitat dynamics in the evolving landscape of the Middle Yangtze River Basin

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Journal of Geographical Sciences 2025年第1期35卷 88-111页

作者： YANG Peng SUN Kaiya zhu Yanchao XIA Jun HUANG Heqing song Jingxia SHI Xiaorui LU Xixi Hubei Key Laboratory of Regional Ecology and Environmental Change School of Geography and Information Engineering China University of Geosciences Department of Geography National University of Singapore State Key Laboratory of Water Resources and Hydropower Engineering Science Wuhan University

Wetlands play a critical role in the global environment. The Middle Yangtze River Basin(MYRB), known for its abundant wetland resources, has experienced notable changes resulting from the complex interplay of environmental factors. Therefore, we investigated the spatiotemporal characteristics of wetland ecological quality in the MYRB from 2001 to *** the random forest(RF) regression algorithm and patch-generated land-use simulation(PLUS) model, we forecasted variations in wetland habitat quality and their determinants under the Shared Socioeconomic Pathway-Representative Concentration Pathway(SSPRCP) framework from 2035 to 2095. The main findings are as follows:(1) The RF algorithm was optimal for land-use and land-cover(LULC) classification in the MYRB from 2001 to 2020,when notable changes were observed in water bodies and buildings. However, the forested area exhibited an increase and decrease of 3.9% and 1.2% under the SSP1-2.6 and SSP5-8.5 scenarios, respectively, whereas farmland exhibited a diminishing trend.(2) Wetlands were primarily concentrated in the central and eastern MYRB, with counties in the southwest exhibiting superior ecological-environmental quality from 2001 to 2020. Notably,wetland coverage revealed significantly high level, significant changes, frequent but relatively minor changes under the SSP1-2.6, SSP2-4.5, and SSP 5-8.5 scenarios, respectively.(3)Regions with lower habitat quality were primarily concentrated in urbanized areas characterized by frequent human activities, indicating a clear degradation in habitat quality across different scenarios. In conclusion, we established a foundational framework for future investigations into the eco-hydrological processes and ecosystem quality of watersheds.

关键词：

Impacts of the annual cycle of background SST in the tropical Pacific on the phase and amplitude of ENSO

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Atmospheric and Oceanic Science Letters 2025年第1期18卷 12-17页

作者： song Jiang Congwen zhu Ning Jiang State Key Laboratory of Severe Weather(LASW) Chinese Academy of Meteorological SciencesBeijingChina College of Earth and Planetary Sciences University of Chinese Academy of SciencesBeijingChina

The dominant annual cycle of sea surface temperature(SST)in the tropical Pacific exhibits an antisymmetric mode,which explains 83.4%total variance,and serves as a background of El Niño-Southern Oscillation(ENSO).However,there is no consensus yet on its anomalous impacts on the phase and amplitude of *** on data during 1982-2022,results show that anomalies of the antisymmetric mode can affect the evolution of ENSO on the interannual scale via Bjerknes feedback,in which the positive(negative)phase of the antisymmetric mode can strengthen El Niño(La Niña)in boreal winter via an earlier(delayed)seasonal cycle transition and larger(smaller)annual *** magnitude of the SST anomalies in the equatorial eastern Pacific can reach more than±0.3◦C,regulated by the changes in the antisymmetric mode based on random sensitivity *** reveal the spatial pattern of the annual cycle associated with the seasonal phase-locking of ENSO evolution and provide new insight into the impact of the annual cycle of background SST on ENSO,which possibly carries important implications for forecasting ENSO.

关键词： Annual cycle SST anomaly Antisymmetric mode ENSO

Deep learning-based end-to-end depth estimation from a single-frame fringe pattern with the FDSUNet++ network

在线全文

学校读者我要写书评

暂无评论

Optoelectronics Letters 2025年

作者： zhu Xinjun ZHANG Jialei song Limei WANG Hongyi TIAN Ruiqin School of Artificial Intelligence， Tiangong University School of Control Science and Engineering， Tiangong University

Recently the depth estimation methods based on deep learning (DL) retains challenging to estimate high-precision in fringe projection structured light three-dimensional (3D) measurement with limited information from a single-frame fringe pattern. In this letter， we proposed a FDSUNet++ convolutional neural network， which consists of a UNet++ base model， an improved squeeze-and-excitation (ISE) block， a Fourier transform (FT) data preprocessing block， and a discrete wavelet transform (DWT) block. The proposed ISE block can improve the ability of feature extraction and the designed FT data preprocessing block preserves the key features of the fringe pattern by Fourier transform. The introduced DWT block reduces the complexity and training cost of the model. By integrating these three blocks into the UNet++ network， it can better achieve depth estimation. Experimental results from two structured light datasets demonstrate that the proposed FDSUNet++ outperforms the state-of-art networks such as UNet， SUNet， R2U-Net， PCTNet and MSUNet++， achieving the best performance in both qualitative and quantitative evaluation.

关键词：

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

竞争环境下双目标零售平台销售模式选择

管理工程学报 2025年第1期39卷 227-242页

作者：张冲朱祎宋君南京邮电大学管理学院江苏南京210003

电子商务行业发展迅速,具有“超越利润”战略导向以实现更广泛目标的组织正变得越来越普遍,兼顾消费者利益成为企业凸显竞争优势的方式。为此,本文构建由一个制造商和两个零售平台组成的两级供应链。其中,制造商和其中一个零售平台追求... 详细信息

电子商务行业发展迅速,具有“超越利润”战略导向以实现更广泛目标的组织正变得越来越普遍,兼顾消费者利益成为企业凸显竞争优势的方式。为此,本文构建由一个制造商和两个零售平台组成的两级供应链。其中,制造商和其中一个零售平台追求自身利润最大化,另外一个零售平台追求自身利润和消费者剩余的线性组合最大化,即考虑双目标最优。本文具体考虑两个零售平台分别选择转售模式或代理模式共四种情形,进而研究双目标零售平台销售模式选择问题,及其对供应链成员最优决策和供应链整体绩效的影响。结果表明:(1)在一定条件下,双目标零售平台能获得比竞争对手更高的利润;(2)当双目标零售平台选择转售模式时,无论其竞争对手选择哪种销售模式,其双目标策略都能使供应链总体利润得到优化;(3)当双目标零售平台的竞争对手选择代理模式时,其自身根据不同情况选择转售模式或者代理模式可以使两个零售平台的销售模式选择形成均衡。

关键词：双目标企业社会责任零售平台销售模式双渠道

A robust statistical prediction model for late-summer heavy precipitation days in North China

在线全文

学校读者我要写书评

暂无评论

Science China Earth Sciences 2025年第1期68卷 158-171页

作者： Shunli JIANG Tingting HAN Xin ZHOU Huijun WANG Zhicong YIN Xiaolei song Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters/Key Laboratory of Meteorological Disaster Ministry of Education Nanjing University for Information Science and Technology Nansen-zhu International Research Centre Institute of Atmospheric Physics Chinese Academy of Sciences Southern Marine Science and Engineering Guangdong Laboratory (zhuhai)

Recently, heavy precipitation(HP) events have occurred frequently in North China(NC), causing devastating economic losses and human fatalities. However, the short-term climate prediction of HP is quite limited. Combining year-to-year increment(DY) method and sliding correlations, we developed a robust seasonal prediction model for late-summer HP days(HPDs) in NC during 1982–2022, utilizing three independent predictors—February sea surface temperature(SST) in the Indian Ocean(SST＿IO), February snow depth over North Asia(SDE＿NA), and May melted snow depth in NC(MSDE＿NC). The SST＿IO anomalies affect NC's precipitation through the Pacific-Japan pattern. The SDE＿NA anomalies are associated with East Asian anomalous anticyclone by southeastern propagation of Rossby wave at Eurasia. The MSDE＿NC anomalies are followed by vertical motion and moisture anomalies in situ and thereby cause precipitation anomalies. This prediction model can well simulate the variations of the HPDs, with a correlation coefficient(CC) of 0.81(0.65) between the observed and predicted HPDs＿DY(HPDs＿anomaly). The percentage with the same sign for 15 extreme HPDs＿anomaly years(PSSE) is 100%.Moreover, in the cross-validation test during 1982–2022, the PSSE for HPDs＿anomaly is as high as 100%, along with a low rootmean-square error of 1.14. For independent hindcasts during 2013–2022, the CC between the observed and predicted HPDs＿DY(HPDs＿anomaly) is 0.93(0.83), together with high Nash-Sutcliffe efficiency(0.82) and agreement index(0.89). Specifically, the predictions are broadly consistent with the observations for 2015, 2016, 2017, 2021, and 2022, reflecting excellent performance of this prediction model of HPDs in NC.

关键词： Heavy precipitation at North China Year-to-year increment approach Robust seasonal prediction Sea surface temperature Snow depth

同方期刊数据库

基于多尺度融合金字塔焦点网络的接触网零部件检测

在线全文

学校读者我要写书评

暂无评论

工程科学学报 2025年第2期47卷 315-327页

作者：朱新宇崔浩锐宋洋中国民用航空飞行学院航空电子电气学院广汉618307 西南交通大学利兹学院成都611756

作为高铁牵引供电系统的重要组成部分,接触网系统承担着向动车组传输电能的重要功能.实际工程运营表明,受弓网交互产生的持续冲击以及外部环境的影响,接触网支撑部件可能会出现“松、脱、断、裂”等缺陷,导致接触网结构可靠性下降,严重... 详细信息

作为高铁牵引供电系统的重要组成部分,接触网系统承担着向动车组传输电能的重要功能.实际工程运营表明,受弓网交互产生的持续冲击以及外部环境的影响,接触网支撑部件可能会出现“松、脱、断、裂”等缺陷,导致接触网结构可靠性下降,严重影响接触网系统稳定运行.因此,及时精确定位接触网支撑部件(CSCs),对保障高铁安全运行和完善接触网检修维护策略具有重大意义.然而,CSCs的检测通常面临着零部件种类多、尺度差异大、部分零部件微小的问题.针对以上问题,本文提出一种基于多尺度融合金字塔焦点网络的接触网零部件检测算法,将平衡模块和特征金字塔模块相结合,提高对小目标的检测性能.首先,设计了可分离残差金字塔聚合模块(SRPAM),用于优化模型多尺度特征提取能力、扩大感受野,缓解CSCs检测的多尺度问题;其次,设计了一种基于平衡特征金字塔的路径聚合网络(PA-BFPN),用于提升跨层特征融合效率和小目标检测性能.最后,通过对比试验、可视化实验和消融实验证明了所提方法的有效性和优越性.其中,所提的MFPFCOS在CSCs数据集上的检测精度(mAP)能够在达到48.6%的同时,实现30的FLOPs(Floating point operations per second),表明所提方法能够在检测精度和检测速度之间保持良好的平衡.

关键词：深度学习目标检测接触网支撑组件(CSCs) 路径聚合特征金字塔(PA-FPN) 空洞空间卷积池化金字塔(ASPP)

Explainable prediction of loan default based on machine learning models

在线全文

学校读者我要写书评

暂无评论

Data Science and Management 2023年第3期6卷 123-133页

作者： Xu zhu Qingyong chu Xinchang song Ping Hu Lu Peng School of Management Wuhan University of TechnologyWuhan430070China Research Institute of Digital Governance and Management Decision Innovation Wuhan University of TechnologyWuhan430070China

Owing to the convenience of online loans,an increasing number of people are borrowing money on online *** the emergence of machine learning technology,predicting loan defaults has become a popular ***,machine learning models have a black-box problem that cannot be *** make the prediction model rules more understandable and thereby increase the user’s faith in the model,an explanatory model must be *** regression,decision tree,XGBoost,and LightGBM models are employed to predict a loan *** prediction results show that LightGBM and XGBoost outperform logistic regression and decision tree models in terms of the predictive *** area under curve for LightGBM is *** accuracies of LightGBM and XGBoost exceed *** precisions of LightGBM and XGBoost exceed ***,we employed the local interpretable model-agnostic explanations approach to undertake an explainable analysis of the prediction *** results show that factors such as the loan term,loan grade,credit rating,and loan amount affect the predicted outcomes.

关键词： Explainable prediction Machine learning Loan default Local interpretable model-agnostic explanations