检索结果-南通市图书馆

FSS: algorithm and neural network accelerator for style transfer

Science China(Information Sciences) 2024年第2期67卷 253-266页

作者： Yi LING Yujie HUANG Yujie CAI Zhaojie LI Mingyu WANG Wenhong LI xiaoyang ZENG State Key Laboratory of ASIC & System Fudan University shanghai explore x technology co. Ltd.

Neural networks(NNs), owing to their impressive performance, have gradually begun to dominate multimedia processing. For resource-co.strained and energy-sensitive mobile devices, an efficient NN accelerator is necessary. Style transfer is an important multimedia application. However, existing arbitrary style transfer networks are co.plex and not well supported by current NN accelerators, limiting their application on mobile devices. Moreover, the quality of style transfer needs improvement. Thus, we design the Fast Style system(FSS), where a novel algorithm and an NN accelerator are proposed for style transfer. In FSS, we first propose a novel arbitrary style transfer algorithm, Fast Style. We propose a light network that co.tributes to high quality and low co.putational co.plexity and a prior mechanism to avoid retraining when the style changes. Then, we redesign an NN accelerator for Fast Style by applying two improvements to the basic NVIDIA deep learning accelerator(NVDLA) architecture. First, a flexible dat FSM and wt FSM are redesigned to enable the original data path to perform other operations(including the GRAM operation)by software programming. Moreover, statistics and judgment logic are designed to utilize the co.tinuity of a video stream and remove the data dependency in the instance normalization, which improves the accelerator performance by 18.6%. The experimental results demonstrate that the proposed Fast Style can achieve higher quality with a lower co.putational co.t, making it more suitable for mobile devices. The proposed NN accelerator is implemented on the xilinx VCU118 FPGA under a 180-MHz clock. Experimental results show that the accelerator can stylize 512×512-pixel video with 20 FPS, and the measured performance reaches up to 306.07 GOPS. The ASIC implementation in TSMC 28 nm achieves about 22 FPS in the case of a 720-p video.

关键词： neural network accelerator style transfer neural network deep learning

来源：

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Research on virtual co.pling technology in rail transit train co.lision protection

引用

Transportation Safety and Environment 2024年第1期6卷 40-46页

作者： Gaosong Li Jinbai Zou Weijie Ma Meng Lan School of Railway Transportation Shanghai Institute of TechnologyShanghai 201418China shanghai Metro First Operation co. LtdShanghai 201418China

In the research on the new generation of intelligent rail transit system,the prevention of train co.lision accidents is still a focus.Virtual co.pling technology co.trols the running speed and interval distance of two or more trains co.peratively through Trainto-Train(T2T)direct co.munication technology.The aim is to form a co.rdinated logical system and jointly co.plete the train co.trol and dispatching.In this paper,a virtual co.pling train formation co.trol method based on the model predictive co.trol(MPC)framework is proposed.From the establishment of the virtual co.pling system,the co.struction of an MPC model and deco.pling method,how to integrate virtual co.pling technology into the rail transit train co.lision protection is studied.The safety and reliability of the system are calculated through fault tree and the method of co.structing reliability function.The research shows that if virtual co.pling technology is used for train co.lision protection,the overall safety and reliability of the train co.trol system is improved.

关键词： virtual co.pling model predictive co.trol co.lision protection

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Automated identification of steel weld defects,a co.volutional neural network improved machine learning approach

引用

Frontiers of Structural and Civil Engineering 2024年第2期18卷 294-308页

作者： Zhan SHU Ao WU Yuning SI Hanlin DONG Dejiang WANG Yifan LI School of Mechanics and Engineering Science Shanghai UniversityShanghai 200444China School of Civil Engineering Shanghai Normal UniversityShanghai 201418China shanghai PinlanData technology co. Ltd.Shanghai 200072China

This paper proposes a machine-learning-based methodology to automatically classify different types of steel weld defects,including lack of the fusion,porosity,slag inclusion,and the qualified(no defects)cases.This methodology solves the shortco.ings of existing detection methods,such as expensive equipment,co.plicated operation and inability to detect internal defects.The study first co.lected percussed data from welded steel members with or without weld defects.Then,three methods,the Mel frequency cepstral co.fficients,short-time Fourier transform(STFT),and co.tinuous wavelet transform were implemented and co.pared to explore the most appropriate features for classification of weld statuses.Classic and co.volutional neural network-enhanced algorithms were used to classify,the extracted features.Furthermore,experiments were designed and performed to validate the proposed method.Results showed that STFT achieved higher accuracies(up to 96.63%on average)in the weld status classification.The co.volutional neural network-enhanced support vector machine(SVM)outperformed six other algorithms with an average accuracy of 95.8%.In addition,random forest and SVM were efficient approaches with a balanced trade-off between the accuracies and the co.putational efforts.

关键词： steel weld machine learning co.volutional neural network weld defect detection classification task percussion

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Research on co.denser Deterioration Evolution Trend Based on ANP-EWM Fusion Health Degree

引用

co.puter Modeling in Engineering & Sciences 2024年第4期139卷 679-698页

作者： Hong Qian Haixin Wang Guangji Wang Qingyun Yan co.lege of Automation Engineering Shanghai University of Electric PowerShanghai200090China shanghai Key Laboratory of Power Station Automation technology Shanghai200072China Empower Information technology(shanghai)co. Ltd.Shanghai201508China China Nuclear Power Engineering Design co. Ltd.Shenzhen518052China

This study presents a proposed method for assessing the co.dition and predicting the future status of co.densers operating in seawater over an extended period.The aim is to address the problems of scaling and co.rosion,which lead to increased loss of co.d resources.The method involves utilising a set of multivariate feature parameters associated with the co.denser as input for evaluation and trend prediction.This methodology offers a precise means of determining the optimal timing for co.denser cleaning,with the ultimate goal of improving its overall performance.The proposed approach involves the integration of the analytic network process with subjective expert experience and the entropy weightmethod with objective big data analysis to develop a fusion health degreemodel.The mathematical model is co.structed quantitatively using the improved Mahalanobis distance.Furthermore,a co.prehensive prediction model is developed by integrating the improved Informer model and Markov error co.rection.This model takes into acco.nt the health status of the equipment and several influencing factors,includingmultivariate feature characteristics.This model facilitates the objective examination and prediction of the progression of equipment deterioration trends.The present study involves the co.putation and verification of the field time series data,which serves to demonstrate the accuracy of the co.denser health-related models proposed in this research.These models effectively depict the real co.dition and temporal variations of the equipment,thus offering a valuable method for determining the precise cleaning time required for the co.denser.

关键词： co.denser health degree improved Mahalanobis distance GC-Informer model Markov error co.rection

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Through-polymer,via technology-enabled,flexible,lightweight,and integrated devices for implantable neural probes

引用

Microsystems & Nanoengineering 2024年第2期10卷 279-291页

作者： Cunkai Zhou Ye Tian Gen Li Yifei Ye Lusha Gao Jiazhi Li Ziwei Liu Haoyang Su Yunxiao Lu Meng Li Zhitao Zhou xiaoling Wei Lunming Qin Tiger H.Tao Liuyang Sun co.lege of Electronics and Information Engineering Shanghai University of Electric PowerShanghaiChina 2020 x-Lab Shanghai Institute of Microsystem and Information TechnologyChinese Academy of SciencesShanghaiChina School of Graduate Study University of Chinese Academy of SciencesBeijingChina State Key Laboratory of Transducer technology Shanghai Institute of Microsystem and Information TechnologyChinese Academy of SciencesShanghaiChina Center of Materials Science and Optoelectronics Engineering University of Chinese Academy of SciencesBeijingChina School of Physical Science and technology ShanghaiTech UniversityShanghaiChina Center for Excellence in Brain Science and Intelligence technology Chinese Academy of SciencesShanghaiChina Neuroxess co. Ltd.(Jiangxi)NanchangJiangxiChina Guangdong Institute of Intelligence Science and technology HengqinZhuhaiGuangdongChina Tianqiao and Chrissy Chen Institute for Translational Research ShanghaiChina

In implantable electrophysiological reco.ding systems,the headstage typically co.prises neural probes that interface with brain tissue and integrated circuit chips for signal processing.While advancements in MEMS and CMOS technology have significantly improved these co.ponents,their interco.nection still relies on co.ventional printed circuit boards and sophisticated adapters.This co.ventional approach adds co.siderable weight and volume to the package,especially for high channel co.nt systems.To address this issue,we developed a through-polymer via(TPV)method inspired by the through-silico. via(TSV)technique in advanced three-dimensional packaging.This innovation enables the vertical integration of flexible probes,amplifier chips,and PCBs,realizing a flexible,lightweight,and integrated device(FLID).The total weight of the FLIDis only 25%that of its co.ventional co.nterparts relying on adapters,which significantly increased the activity levels of animals wearing the FLIDs to nearly match the levels of co.trol animals without implants.Furthermore,by inco.porating a platinum-iridium alloy as the top layer material for electrical co.tact,the FLID realizes exceptional electrical performance,enabling in vivo measurements of both local field potentials and individual neuron action potentials.These findings showcase the potential of FLIDs in scaling up implantable neural reco.ding systems and mark a significant advancement in the field of neurotechnology.

关键词： alloy neural reco.ding

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

On the electric behaviour of co.ductive grease inside the co.tact zone

引用

Friction 2024年第4期12卷 683-697页

作者： Yuxin ZHOU Jinjie LIU xinming LI He CHONG Bo HAN Qinghua BAI SKF(shanghai)Automotive Technologies co. Ltd.Shanghai 201814China School of Mechanical and Automotive Engineering Qingdao University of TechnologyQingdao 266250China

co.ductive grease has been co.sidered as a potential solution in addressing electric erosion issue inside motor bearings.Understanding the co.ductive grease performance,especially the electric and tribological behaviour is crucial for its application.This paper co.bines bulk grease characterization and co.tact film study co.sidering both electrical and tribological loading co.ditions.The results show that the electric performance for the selected co.ductive grease sample is electric field dependent,including both frequency and voltage amplitude.Its film impedance inside the co.tact area is closely related to the external electric field,running co.dition,and lubrication regime.The following film impedance calculation tells that not only bulk grease electric parameters,but the co.tact parameter,lubricant film co.tour and lubrication co.dition should be co.sidered when performing the overall bearing electrical modelling.

关键词： Electric erosion lubricant co.ductivity grease film impedance

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Reversible Mn^(2+)/Mn^(4+)double-electron redox in P3-type layer-structured sodium-ion cathode

引用

Journal of Energy Chemistry 2024年第2期89卷 79-88,I0004页

作者： Jie Zeng Jian Bao Ya Zhang xun-Lu Li Cui Ma Rui-Jie Luo Chong-Yu Du xuan xu Zhe Mei Zhe Qian Yong-Ning Zhou Department of Materials Science Fudan UniversityShanghai 200433China shanghai xuanyi New Energy Development co. LtdShanghai 201800China Global Institute of Future technology Shanghai Jiao Tong UniversityShanghai 200240China

The balance between cationic redox and oxygen redox in layer-structured cathode materials is an important issue for sodium batteries to obtain high energy density and co.siderable cycle stability.Oxygen redox can co.tribute extra capacity to increase energy density,but results in lattice instability and capacity fading caused by lattice oxygen gliding and oxygen release.In this work,reversible Mn^(2+)/Mn^(4+)redox is realized in a P3-Na_(0.65)Li_(0.2)co.(0.05)Mn_(0.75)O_(2)cathode material with high specific capacity and structure stability via co.substitution.The co.tribution of oxygen redox is suppressed significantly by reversible Mn^(2+)/Mn^(4+)redox without sacrificing capacity,thus reducing lattice oxygen release and improving the structure stability.Synchrotron x-ray techniques reveal that P3 phase is well maintained in a wide voltage window of 1.5-4.5 V vs.Na^(+)/Na even at 10 C and after long-term cycling.It is disclosed that charge co.pensation from co.Mn-ions co.tributes to the voltage region below 4.2 V and O-ions co.tribute to the whole voltage range.The synergistic co.tributions of Mn^(2+)/Mn^(4+),co.(2+)/co.(3+),and O^(2-)/(O_n)^(2-)redox in P3-Na_(0.65)Li_(0.2)co.(0.05)Mn_(0.75)O_(2)lead to a high reversible capacity of 215.0 m A h g^(-1)at 0.1 C with co.siderable cycle stability.The strategy opens up new opportunities for the design of high capacity cathode materials for rechargeable batteries.

关键词： Sodium batteries Cathode materials Layered structure co.substitution

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Differential interference co.trast phase edging net:an all-optical learning system for edge detection of phase objects

引用

Chinese Optics Letters 2024年第1期22卷 21-27页

作者：李一鸣李然陈泉栾海涛卢海军杨晖顾敏张启明 Institute of Photonic Chips University of Shanghai for Science and TechnologyShanghai 200093China School of Optical-Electrical and co.puter Engineering University of Shanghai for Science and TechnologyShanghai 200093China Nokia shanghai Bell co. Ltd.Shanghai 201206China co.lege of Medical Instruments Shanghai University of Medicine and Health SciencesShanghai 201318China

Edge detection for low-co.trast phase objects cannot be performed directly by the spatial difference of intensity distribution.In this work,an all-optical diffractive neural network(DPENet)based on the differential interference co.trast principle to detect the edges of phase objects in an all-optical manner is proposed.Edge information is enco.ed into an interference light field by dual Wollaston prisms without lenses and light-speed processed by the diffractive neural network to obtain the scale-adjustable edges.Simulation results show that DPENet achieves F-sco.es of 0.9308(MNIST)and 0.9352(NIST)and enables real-time edge detection of biological cells,achieving an F-sco.e of 0.7462.

关键词： diffractive neural network edge detection phase objects

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Polarized imaging dynamic light scattering for simultaneous measurement of nanoparticle size and morphology

引用

Particuology 2024年第2期85卷 213-223页

作者： Bingyao Wang Tianyi Cai xiaoshu Cai Wu Zhou Zeqi Liu Schoolof Energy and Power Engineering University of Shanghai for Science and TechnologyShanghai200093China shanghai Key Laboratory of Multiphase Flow and Heat Transfer in Power Engineering Shanghai200093China Jiaxing MeaParTech Instrument technology co. LtdZhejiang314000China

The performance of nanoparticles is often affected by particle size and morphology.Currently,electron microsco.y or atomic force microsco.y is typically utilized to determine the size and morphology of nanoparticles.However,there are issues such as difficult sample preparation,long processing times,and challenges in quantitative characterization.Therefore,it is of great significance to develop a fast,accu-rate,and statistical method to measure the size and morphology of nanoparticles.In this study,a new method,called polarized imaging dynamic light scattering(PIDLS),is proposed.The nanoparticles are irradiated with a vertical linearly polarized laser beam,and a polarization camera co.lected the dynamic light scattering images of particles at four different polarization directions(0°,45°,90°,and 135°)at a scattering angle of 90°.The average particle size and distribution are obtained using the imaging dy-namic light scattering method at 0°polarization direction,and the morphology of the particles is ob-tained based on the depolarization patterns of the scattered light.The optical sphericityΦis defined based on the degree of linear polarization(DoLP).It is also implemented for the quantitative evaluation of the sphericity of the nanoparticles,including spherical,octahedral,nanoplate,nanorod,and linear ones.Together with the Poincarésphere parameterψ,the morphology of the nanoparticles can be roughly identified.In addition,PIDLS enables the measurement of particle size and morphology distributions simultaneously for evaluating the uniformity of particles.The effectiveness of PIDLS is verified by the measurement of five kinds of industrial titanium dioxide as well.

关键词： Imaging dynamic light scattering Polarized light scattering Non-spherical nanoparticles Morphology Online measurement

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Scenarios analysis and performance assessment of blockchain integrated in 6G scenarios

引用

Science China(Information Sciences) 2024年第7期67卷 5-26页

作者： Bo LI Guanjie CHENG Honghao GAO xueqiang YAN Shuiguang DENG co.lege of co.puter Science and technology Zhejiang University co.lege of co.puter Science and technology Shanghai University 2012 Lab Huawei Technologies Co. Ltd.

Emerging applications such as smart city infrastructures and virtual reality landscapes are setting rigorous benchmarks for 6G mobile networks, requiring elevated levels of co.fidentiality, integrity, nonrepudiation, authentication, and stringent access co.trols. Blockchain technology is heralded as a transformative enabler for meeting 6G standards, owing to its intrinsic attributes. However, a gap exists in the holistic investigation of blockchain's applicability in 6G realms, particularly addressing the “whether”, “when”, and“how” of its deployment. Present research trails in developing robust methodologies to gauge blockchain's efficacy within 6G use cases. Addressing this, our study introduces a novel co.fluence of blockchain with 6G networks, where data resides in distributed Hash tables(DHTs) while their hashes are secured in distributed ledger technology(DLT), harnessing blockchain's co.e strengths-immutability, traceability, and fortified security. We delineate seven specific 6G use cases poised for enhancement through blockchain integration,and scrutinize the rationale, nature, and timing of this co.vergence. Furthermore, we devise a co.prehensive methodology for assessing blockchain's performance metrics and scalability in 6G environments. Our extensive experimental analyses evaluate the synergistic performance of this integration, revealing that the Quorum blockchain satisfactorily supports 80% of 6G scenarios. The findings suggest that, with appropriate co.figurations, co.sortium blockchains are well-equipped to fulfill the demanding performance and scalability requisites of 6G networks.

关键词： blockchain distributed ledger technology (DLT) 6G performance assessment 6G scenarios analysis transaction arrival rate

来源：

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

欢迎您,

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：