检索结果-内蒙古大学图书馆

Dynamic resolution network 21

学校读者我要写书评

暂无评论

Dynamic resolution network

Proceedings of the 35th International Conference on Neural Information Processing Systems

作者： Mingjian Zhu Kai Han Enhua Wu Qiulin Zhang Ying Nie Zhenzhong Lan Yunhe Wang Zhejiang University and Huawei Noah's Ark Lab. and School of Engineering Westlake University and Institute of Advanced Technology Westlake Institute for Advanced Study Huawei Noah's Ark Lab. and State Key Lab of Computer Science ISCAS & University of Chinese Academy of Sciences State Key Lab of Computer Science ISCAS & University of Chinese Academy of Sciences and University of Macau BUPT Huawei Noah's Ark Lab. School of Engineering Westlake University and Institute of Advanced Technology Westlake Institute for Advanced Study

ISBN: (纸本)9781713845393

Deep convolutional neural networks (CNNs) are often of sophisticated design with numerous learnable parameters for the accuracy reason. To alleviate the expensive costs of deploying them on mobile devices, recent works have made huge efforts for excavating redundancy in pre-defined architectures. Nevertheless, the redundancy on the input resolution of modern CNNs has not been fully investigated, i.e., the resolution of input image is fixed. In this paper, we observe that the smallest resolution for accurately predicting the given image is different using the same neural network. To this end, we propose a novel dynamic-resolution network (DRNet) in which the input resolution is determined dynamically based on each input sample. Wherein, a resolution predictor with negligible computational costs is explored and optimized jointly with the desired network. Specifically, the predictor learns the smallest resolution that can retain and even exceed the original recognition accuracy for each image. During the inference, each input image will be resized to its predicted resolution for minimizing the overall computation burden. We then conduct extensive experiments on several benchmark networks and datasets. The results show that our DRNet can be embedded in any off-the-shelf network architecture to obtain a considerable reduction in computational complexity. For instance, DR-ResNet-50 achieves similar performance with an about 34% computation reduction, while gaining 1.4% accuracy increase with 10% computation reduction compared to the original ResNet-50 on ImageNet.

关键词：

High-Performance Sub-Micron Cigsse Solar Cells Optimized for Sodium Doping by Adjusting Diffusion Barriers

学校读者我要写书评

暂无评论

SSRN 2022年

作者： Wang, C. Lia, Y. Zhuang, Daming Zhao, Ming Tong, H. Wang, H. Wei, J. Gong, Q. School of Materials Science and Engineering Tsinghua University Beijing100084 China Key Lab. for Advanced Materials Processing Technology Ministry of Education State Key Lab. of New Ceramics and Fine Processing College of Mechanical and Vehicle Engineering Taiyuan University of Technology Taiyuan030024 China

Alkali metal Na is beneficial to the performance of CuInGaSSe thin film solar cells. This work studies the addition and optimization of a sputtering SiO 2 barrier layer on the SLG to optimize the Na content that diffuses into the absorber layer from the SLG. The resulting Se content and distribution, the influence and change of the MoSe 2 thickness are studied. Also, the changes in quality of the p-n junction and the defects of the absorption layer are studied. Last, the resulting carrier separation and changes in the carrier recombination rate of each part of the absorber are analyzed. For sub-micron CIGS devices, it is clear that untreated SLG provides excessive amounts of Na. After optimization and analysis of the barrier layer thickness, the average conversion efficiency is increased by 34% compared with the sub-optimal sample, and the efficiency reaches 18.9%. © 2022, The Authors. All rights reserved.

关键词： Efficiency

Recognition of electromagnetic Signals Based on the Spiking Convolutional Neural Network

学校读者我要写书评

暂无评论

Wireless Communications and Mobile Computing 2022年第0期2022卷

作者： Tao, Shifei Xiao, Shuai Gong, Shuaige Wang, Huifu Ding, Hao Wang, Hao State Key Lab. of Complex Electromagnetic Environment Effects on Electronics and Information System Luoyang471003 China School of Electrical and Optical Engineering Nanjing University of Science and Technology Nanjing210094 China Engineering Training Center Nanjing University of Science and Technology Nanjing210094 China

Feature extraction and recognition of signals are the bases of a cognitive radio. Traditional manual extraction for signals' features becomes difficult in the complex electromagnetic environment. Although convolutional neural networks (CNNs) can extract signal features automatically, they have low accuracy in recognizing electromagnetic signals at low signal-to-noise ratios (SNRs) due to the agility of signals. Considering the great potential of spiking neural networks (SNNs) in classification, a spiking convolution neural network (SCNN) for the recognition of electromagnetic signals is proposed in this paper. The SCNN effectively integrates the extraction ability of spatial features in CNNs and temporal features in SNNs. Since the SCNN is difficult to train, the strategy of surrogate gradient is proposed to train it. By taking the 2-dimensional time-frequency distribution of 6 signals as input, the SCNN can effectively identify different signals at low SNRs. The method proposed in this paper contributes to promote the research and application of SNNs in the recognition of electromagnetic signals. © 2022 Shifei Tao et al.

关键词： Extraction

Investigations of Sodium Droplet Spreading and Evaporation on Surfaces with Different Wettability Based on Lattice Boltzmann Simulations

学校读者我要写书评

暂无评论

SSRN 2024年

作者： Tang, Songsheng Guo, Kailun Tian, Zhixing Wang, Mingjun Wang, Chenglong He, Xiaoqiang Su, Guanghui Qiu, Suizheng Tian, Wenxi School of Nuclear Science and Technology State Key Lab. of Multiphase Flow in Power Engineering Shaanxi Key Lab. of Advanced Nuclear Energy and Technology Xi’an Jiaotong University Shaanxi Xi’an710049 China Key Lab. of Nuclear Reactor System Design Technology China Nuclear Power Research and Design Institute Chengdu610213 China

The spreading and evaporation of sodium droplets is one of the fundamental phenomena in high-temperature heat pipes. The LBM sodium two-phase flow and phase transition model, which can simulate a huge liquid-vapor density ratio, is first established and validated. Then, the spreading and evaporation processes of sodium droplets on solid walls with different wettability are numerically simulated and analyzed. The main conclusions obtained are as follows: (1) The dimensionless spreading diameter and time of the droplet spreading process conforms to Tanner's power law, and the power exponent has exponential relationship with the equilibrium wetting angle;(2) The maximum dimensionless spreading diameter is negatively and linearly correlated with the equilibrium wetting angle;(3) Evaporation of sodium droplets under constant heat flow conditions shows a mixed pattern. The droplets will be detached from the droplet wall at the late stage of evaporation, and the larger the equilibrium wetting angle the sooner the droplets detach. © 2024, The Authors. All rights reserved.

关键词： Evaporation

Omnidirectional Conformal Cloak Without Geometrical Dispersion

学校读者我要写书评

暂无评论

Physical Review Applied 2019年第6期12卷 064009-064009页

作者： Yichao Liu Fei Sun Sailing He Key Lab of Advanced Transducers and Intelligent Control System Ministry of Education and Shanxi Province College of Physics and Optoelectronics Taiyuan University of Technology Taiyuan 030024 China State Key Laboratory of Modern Optical Instrumentations Centre for Optical and Electromagnetic Research JORCEP East Building #5 Zijingang Campus Zhejiang University Hangzhou 310058 China Department of Electromagnetic Engineering School of Electrical Engineering Royal Institute of Technology (KTH) S-100 44 Stockholm Sweden

The phase delay problem of invisibility cloaks designed by optical conformal mapping is referred to the optical path difference between the waves in upper and lower Riemann sheets. As the frequency changes, the phase delay varies accordingly (here, we call it geometrical dispersion), which restricts the cloaks to working only at some discrete frequencies. Here, we solve the phase-delay problem by placing an optically null medium in the lower Riemann sheet. The conformal cloaks designed by using our method have many advantages, such as a continuous working band (no geometrical dispersion), omnidirectionality, robustness under frequency shifts, and no need for any complex materials that are both anisotropic and inhomogeneous. We also design metallic microchannels to realize our cloaks.

关键词： Optical cloaking Devices Optical materials & elements Optical conformal mapping

Design and analysis of a self-tuning proportional and integral controller for active queue management routers to support TCP flows

学校读者我要写书评

暂无评论

Design and analysis of a self-tuning proportional and integr...

INFOCOM 2006: 25th IEEE International Conference on Computer Communications

作者： Xiong, Naixue Défago, Xavier Jia, Xiaohua Yan, G. He, Yanxiang School of Information Science Japan Advanced Institute of Science and Technology State Key Lab. of Software Engineering Computer School Wuhan University School of Information Science Japan Advanced Institute of Science and Technology 1-1 Asahidai Nomi Ishikawa 923-1292 Japan Kowloon Hong Kong IEEE IEEE Communication Society Department of Computer Science Wuhan University City University of Hong Kong Kowloon Hong Kong School of Computer Science and Technology Wuhan University of Science and Technology China SMIEEE

ISBN: (纸本)1424402212

Active Queue Management (AQM) is an effective method used in Internet routers for congestion control, and to achieve a tradeoff between link utilization and delay. The de facto standard, the Random Early Detection (RED) AQM scheme, and most of its variants use average queue length as a congestion indicator to trigger packet dropping. This paper proposes a novel AQM algorithm, called Self-tuning Proportional and Integral RED (SPI-RED), as an extension of RED. SPI-RED is based on a Self-tuning Proportional and Integral controller, which not only considers the average queue length at the current time point, but also takes into consideration the past average queue lengths during a round-trip time. Furthermore, we give theoretical analysis of the system stability and give guidelines for the selection of feedback gains for the TCP/RED system to stabilize the average queue length at a desirable level. Extensive simulations have been conducted with ns2. The simulation results have demonstrated that the proposed SPI-RED algorithm outperforms the existing AQM schemes in terms of drop probability and stability.

关键词： Queueing networks

Development and Verification of Safety Analysis Code to Model and Simulate Steam Power Conversion System in Sodium-Cooled Fast Reactor

学校读者我要写书评

暂无评论

SSRN 2023年

作者： Liu, Shuo Jin, Zhao Wang, Chenglong Zhang, Dalin Tian, Wenxi Qiu, Suizheng Su, Guanghui Yang, Jun School of Nuclear Science and Technology State Key Lab. of Multiphase Flow in Power Engineering Shaanxi Key Lab. of Advanced Nuclear Energy and Technology Xi’an Jiaotong University Shaanxi Xi’an710049 China China Institute of Atomic Energy Beijing102413 China

Sodium-cooled fast reactor (SFR) is one of the most promising generation IV reactor type. SFR system is composed of three heat transfer circuits. SFR will generate electricity, which requires accuracy prediction of steam power conversion system. Therefore, a one-dimensional thermal-hydraulic analysis code of the steam power conversion system of SFR has been developed. A modular modeling method was applied. Each equipment is packaged as a separate module. The system model can be obtained by connecting modules to achieve flexible modeling. Compared with RELAP5, the code uses a general algorithm to solve ODE, which enhances the universality of code. The model is verified with design results of Daya Bay Nuclear Power Plant and BN-600 steam generator. The maximum relative errors for pressure and temperature are 5.44% and 2.40%, respectively. It is proven that the code developed can be used in the analysis and design the steam power conversion system. The steady-state analysis of SFR steam power conversion circuit is carried out. The design could provide a reference for the future development of steam power conversion system in SFR. © 2023, The Authors. All rights reserved.

关键词： Sodium-cooled fast reactors

Dynamic Modeling of Steam Power Conversion System in Sodium-Cooled Fast Reactor

学校读者我要写书评

暂无评论

SSRN 2023年

作者： Liu, Shuo Jin, Zhao Wang, Chenglong Zhang, Dalin Tian, Wenxi Qiu, Suizheng Su, G.H. Yang, Jun School of Nuclear Science and Technology State Key Lab. of Multiphase Flow in Power Engineering Shaanxi Key Lab. of Advanced Nuclear Energy and Technology Xi’an Jiaotong University Shaanxi Xi’an710049 China China Institute of Atomic Energy Beijing102413 China

Sodium-cooled fast reactor (SFR) is the most promising generation IV reactor type. SFR systems are composed of three heat transfer circuits. Currently, research on steam power conversion system is scarce because of complex structure. Therefore, a one-dimensional thermal-hydraulic analysis code of the steam power conversion system of SFR has been developed. Models of main components have been established, including once-through steam generator (OTSG), turbine, deaerator, feedwater heater, condenser, pump and moisture separator reheater (MSR). The model is verified with design results of Daya Bay Nuclear Power Plant and BN-600 steam generator. The maximum relative errors for pressure and temperature are 5.44% and 2.40%, respectively. It is proven that the code developed can analysis and design the steam power conversion system. The steady-state analysis of SFR steam power conversion circuit is carried out. The design could provide a reference for the future development of steam power conversion system in SFR. © 2023, The Authors. All rights reserved.

关键词： Sodium-cooled fast reactors

engineering-oriented investigation of magnetic property modeling and application

学校读者我要写书评

暂无评论

Engineering-oriented investigation of magnetic property mode...

作者： Cheng, Zhiguang Forghani, Behzad Wang, Xuegang Liu, Lanrong Liu, Tao Fan, Yana Zhang, Junjie Zhao, Xiaojun Liu, Yang Institute of Power Transmission and Transformation Technology Baobian Electric Co. Ltd Baoding Hebei071056 China Infolytica Corporation Place du Parc MontrealQC Canada Department of Electrical Engineering North China Electric Power University Baoding Hebei China State Key Lab. of Advanced Transmission Technology Global Energy Interconnection Research Institute Beijing China

This paper discusses engineering-oriented magnetic material property modeling, fully taking into account the requirements posed by industrial applications, involving the magnetic properties measured under both standard and extreme conditions, the combination of material modeling and numerical analysis, and the validation of modeling and simulation. It presents some application notes on typical magnetic measurement and modeling, concerning the Epstein combination and data weighted processing method, the additional iron loss caused by 3-D excitation, the examination of magnetic flux inside laminated stacks, and the effect of both B-H curves and solvers on loss analysis, based on the authors' recent research. In addition, future research projects on the determination and application of the magnetic material properties, under practical working conditions, are proposed. © 2017 - IOS Press and the authors. All rights reserved.

关键词： Magnetic variables measurement