检索结果-南通市图书馆

Research on a deconvolution algorithm for laser-induced fluorescence diagnosis based on the maximum entropy principle

在线全文

学校读者我要写书评

暂无评论

Plasma Science and Technology 2024年第7期26卷 93-107页

作者：雷清雲杨雄程谋森张帆郭大伟李小康肖文杰 National University of defense Technology Changsha 410073People’s Republic of China Bejing institute of Tracking and Telecommunications Technology Beijing 100094People’s Republic of China

Laser-induced fluorescence(LIF)spectroscopy is employed for plasma diagnosis,necessitating the utilization of deconvolution algorithms to isolate the Doppler effect from the raw spectral ***,direct deconvolution becomes invalid in the presence of noise as it leads to infinite amplification of high-frequency noise *** address this issue,we propose a deconvolution algorithm based on the maximum entropy *** validate the effectiveness of the proposed algorithm by utilizing simulated LIF spectra at various noise levels(signal-to-noise ratio,SNR=20–80 d B)and measured LIF spectra with Xe as the working *** the typical measured spectrum(SNR=26.23 d B)experiment,compared with the Gaussian filter and the Richardson–Lucy(R-L)algorithm,the proposed algorithm demonstrates an increase in SNR of 1.39 d B and 4.66 d B,respectively,along with a reduction in the root-meansquare error(RMSE)of 35%and 64%,***,there is a decrease in the spectral angle(SA)of 0.05 and 0.11,*** the high-quality spectrum(SNR=43.96 d B)experiment,the results show that the running time of the proposed algorithm is reduced by about98%compared with the R-L iterative ***,the maximum entropy algorithm avoids parameter optimization settings and is more suitable for automatic *** conclusion,the proposed algorithm can accurately resolve Doppler spectrum details while effectively suppressing noise,thus highlighting its advantage in LIF spectral deconvolution applications.

关键词： maximum entropy spectral deconvolution laser-induced fluorescence spectrum

Learning to Branch in Combinatorial Optimization With Graph Pointer Networks

在线全文

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2024年第1期11卷 157-169页

作者： Rui Wang Zhiming Zhou Kaiwen Li Tao Zhang Ling Wang Xin Xu Xiangke Liao IEEE the Xiangjiang Laboratory and the College of Systems Engineering National University of Defense TechnologyChangsha 410073China the institute of Automation Chinese Academy of SciencesBeijing 100190China the College of Systems Engineering National University of Defense TechnologyChangsha 410073 the Hunan Key Laboratory of Multi-Energy System Intelligent Interconnection Technology Changsha 410073China the Department of Automation Tsinghua UniversityBeijing 100084China the College of Intelligence Science and Technology National University of Defense TechnologyChangsha 410073China the College of Computer Science and Technology National University of Defense TechnologyChangsha 410073China

Traditional expert-designed branching rules in branch-and-bound(B&B) are static, often failing to adapt to diverse and evolving problem instances. Crafting these rules is labor-intensive, and may not scale well with complex *** the frequent need to solve varied combinatorial optimization problems, leveraging statistical learning to auto-tune B&B algorithms for specific problem classes becomes attractive. This paper proposes a graph pointer network model to learn the branch rules. Graph features, global features and historical features are designated to represent the solver state. The graph neural network processes graph features, while the pointer mechanism assimilates the global and historical features to finally determine the variable on which to branch. The model is trained to imitate the expert strong branching rule by a tailored top-k Kullback-Leibler divergence loss function. Experiments on a series of benchmark problems demonstrate that the proposed approach significantly outperforms the widely used expert-designed branching rules. It also outperforms state-of-the-art machine-learning-based branch-and-bound methods in terms of solving speed and search tree size on all the test instances. In addition, the model can generalize to unseen instances and scale to larger instances.

关键词： Branch-and-bound(B&B) combinatorial optimization deep learning graph neural network imitation learning

Experimental and numerical study on protective effect of RC blast wall against air shock wave

在线全文

学校读者我要写书评

暂无评论

Defence Technology（防务技术） 2024年第1期31卷 567-579页

作者： Xin-zhe Nian Quan-min Xie Xin-li Kong Ying-kang Yao Kui Huang State Key Laboratory of Precision Blasting Jianghan UniversityWuhan430056China Hubei Key Laboratory of Blasting Engineering Jianghan UniversityWuhan430056China Hubei(Wuhan)institute of Explosion and Blasting Technology Jianghan UniversityWuhan430056China College of defense Engineering Army Engineering University of PLANanjing210007China institute of defense Engineering AMSPLALuoyang471023China

Prototype experiments were carried out on the explosion-proof performance of the RC blast *** mass of TNT detonated in the experiments is 5 kg and 20 kg *** shock wave overpressure was tested in different *** above experiments were numerically simulated,and the simulated shock wave overpressure waveforms were compared with that tested and given by CONWEP *** results show that the numerically simulated waveform is slightly different from the test waveform,but similar to CONWEP *** dimensional analysis and numerical simulation under different working conditions,the equation for the attenuation rate of the diffraction overpressure behind the blast wall was *** to the corresponding standards,the degree of casualties and the damage degree of the brick concrete building at a certain distance behind the wall can be determined when parameters are *** above results can provide a reference for the design and construction of the reinforced concrete blast wall.

关键词： Blast wall Shock wave Diffraction Overpressure Protection

Dynamic Update Scheme of Spectrum Information Based on Spectrum Opportunity Incentive in the Database-Assisted Dynamic Spectrum Management

在线全文

学校读者我要写书评

暂无评论

China Communications 2024年第3期21卷 75-90页

作者： Zhang Yu Chen Yong He Panfeng Cai Yueming Sixty-third Research institute National University of Defense TechnologyNanjing 210007China College of Communications Engineering Army of Engineering UniversityNanjing 210007China

To solve the problem of delayed update of spectrum information(SI) in the database assisted dynamic spectrum management(DB-DSM), this paper studies a novel dynamic update scheme of SI in DB-DSM. Firstly, a dynamic update mechanism of SI based on spectrum opportunity incentive is established, in which spectrum users are encouraged to actively assist the database to update SI in real time. Secondly, the information update contribution(IUC) of spectrum opportunity is defined to describe the cost of accessing spectrum opportunity for heterogeneous spectrum users, and the profit of SI update obtained by the database from spectrum allocation. The process that the database determines the IUC of spectrum opportunity and spectrum user selects spectrum opportunity is mapped to a Hotelling model. Thirdly, the process of determining the IUC of spectrum opportunities is further modelled as a Stackelberg game by establishing multiple virtual spectrum resource providers(VSRPs) in the database. It is proved that there is a Nash Equilibrium in the game of determining the IUC of spectrum opportunities by VSRPs. Finally, an algorithm of determining the IUC based on a genetic algorithm is designed to achieve the optimal IUC. The-oretical analysis and simulation results show that the proposed method can quickly find the optimal solution of the IUC, and ensure that the spectrum resource provider can obtain the optimal profit of SI update.

关键词： database-assisted dynamic spectrum management Hotelling information updating contribution spectrum information updating Stackelberg

The quantum Ising model for perfect matching and solving it with variational quantum eigensolver

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第9期67卷 306-320页

作者： Qilin ZHENG Pingyu ZHU Chao WU Miaomiao YU Weihong LUO Ping XU institute for Quantum Information and State Key Laboratory of High Performance Computing College of Computer Science and Technology National University of Defense Technology Hefei National Laboratory

Obtaining all perfect matchings of a graph is a tough problem in graph theory, and its complexity belongs to the #P-Complete class. The problem is closely related to combinatorics, marriage matching problems, dense subgraphs, the Gaussian boson sampling, chemical molecular structures, and dimer *** this paper, we propose a quadratic unconstrained binary optimization formula of the perfect matching problem and translate it into the quantum Ising model. We can obtain all perfect matchings by mapping them to the ground state of the quantum Ising Hamiltonian and solving it with the variational quantum eigensolver. Adjusting the model's parameters can also achieve the maximum or minimum weighted perfect matching. The experimental results on a superconducting quantum computer of the Origin Quantum Computing Technology Company show that our model can encode 2~n dimensional optimization space with only O（n） qubits consumption and achieve a high success probability of the ground state corresponding to all perfect matchings. In addition, the further simulation results show that the model can support a scale of more than 14 qubits, effectively resist the adverse effects of noise, and obtain a high success probability at a shallow variational depth. This method can be extended to other combinatorial optimization problems.

关键词： perfect matching Ising model quantum Hamiltonian variational quantum eigensolver quadratic unconstrained binary optimization

Autonomous relative optical navigation based on unified modeling of external systematic errors

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Aeronautics 2024年第3期37卷 282-289页

作者： Dayi WANG Bowen SUN Maodeng LI Jiaxing LI Haiyin ZHOU Beijing institute of Spacecraft System Engineering Beijing 100191China College of Sciences National University of Defense TechnologyChangsha 410073China National Key Laboratory of Space Intelligent Control Beijing Institute of Control EngineeringBeijing 100191China

To solve the problem that external systematic errors of the optical camera cannot be fully estimated due to limited computing resources,a unified dimensionality reduction representation method for the external systematic errors of the optical camera is proposed,and autonomous relative optical navigation is *** camera translational and misalignment errors are converted into a three-dimensional rotation error,whose differential model can be established through specific attitude control and appropriate ***,the rotation error and the relative motion state are jointly estimated in an augmented Kalman filter *** with the traditional method that estimates the camera translational and misalignment errors,the proposed method reduces the computational complexity in that the estimated state dimension is ***,as demonstrated by numerical simulation,the estimation accuracy is improved significantly.

关键词： Unified model Systematic errors Optical navigation Observability analysis Calibration

Simulating a Chern Insulator with𝐶C=±2 on Synthetic Floquet Lattice

在线全文

学校读者我要写书评

暂无评论

Chinese Physics Letters 2024年第9期41卷 1-8页

作者： Ling-Xiao Lei Wei-Chen Wang Guang-Yao Huang Shun Hu Xi Cao Xin-Fang Zhang Ming-Tang Deng Ping-Xing Chen institute for Quantum Science and Technology College of ScienceNational University of Defense TechnologyChangsha 410073China institute for Quantum Information&State Key Laboratory of High Performance Computing College of Computer Science and TechnologyNational University of Defense TechnologyChangsha 410073China Greatwall Quantum Laboratory Changsha 410006China Hefei National Laboratory Hefei 230088China

The synthetic Floquet lattice,generated by multiple strong drives with mutually incommensurate frequencies,provides a powerful platform for quantum simulation of topological *** this study,we propose a 4-band tight-binding model of the Chern insulator with a Chern number𝐶=±2 by coupling two layers of the half Bernevig–Hughes–Zhang lattice and subsequently mapping it onto the Floquet lattice to simulate its topological *** determine the Chern number of our Floquet-version model,we extend the energy pumping method proposed by Martin et al.[2017 ***.X 7041008]and the topological oscillation method introduced by Boyers et al.[2020 ***.125160505],followed by numerical simulations for both *** simulation results demonstrate the successful extraction of the Chern number using either of these methods,providing an excellent prediction of the phase diagram that closely aligns with the theoretical one derived from the original bilayer half Bernevig–Hughes–Zhang ***,we briefly discuss a potential experimental implementation for our *** work demonstrates significant potential for simulating complex topological matter using quantum computing platforms,thereby paving the way for constructing a more universal simulator for non-interacting topological quantum states and advancing our understanding of these intriguing phenomena.

关键词： properties. topological quantum

Memory-enhanced text style transfer with dynamic style learning and calibration

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

Science China(Information Sciences) 2024年第4期67卷 181-196页

作者： Fuqiang LIN Yiping SONG Zhiliang TIAN Wangqun CHEN Diwen DONG Bo LIU College of Computer Science and Technology National University of Defense Technology Department of Computer Science and Engineering The Hong Kong University of Science and Technology Strategic Assessments and Consultation institute Academy of Military Sciences

Text style transfer aims to rephrase a sentence to match the desired style while retaining the original content. As a controllable text generation task, mainstream approaches use content-independent style embedding as control variables to guide stylistic generation. Nonetheless, stylistic properties are contextsensitive even under the same style. For example, “delicious” and “helpful” convey positive sentiments,although they are more likely to describe food and people, respectively. Therefore, desired style signals must vary with the content. To this end, we propose a memory-enhanced transfer method, which learns fine-grained style representation concerning content to assist transfer. Rather than employing static style embedding or latent variables, our method abstracts linguistic characteristics from training corpora and memorizes subdivided content with the corresponding style representations. The style signal is dynamically retrieved from memory using the content as a query, providing a more expressive and flexible latent style space. To address the imbalance between quantity and quality in different content, we further introduce a calibration method to augment memory construction by modeling the relationship between candidate *** results obtained using three benchmark datasets confirm the superior performance of our model compared to competitive approaches. The evaluation metrics and case study also indicate that our model can generate diverse stylistic phrases matching context.

关键词： style transfer memory-enhanced method text generation deep learning text representation

OpBench: an operator-level GPU benchmark for deep learning

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第9期67卷 76-87页

作者： Qingwen GU Bo FAN Zhengning LIU Kaicheng CAO Songhai ZHANG Shimin HU Department of Computer Science and Technology Tsinghua University National Innovation institute of defense Technology Academy of Military Science Beijing Fitten Technology Co. Ltd.

Operators(such as Conv and ReLU) play an important role in deep neural networks. Every neural network is composed of a series of differentiable operators. However, existing AI benchmarks mainly focus on accessing model training and inference performance of deep learning systems on specific models. To help GPU hardware find computing bottlenecks and intuitively evaluate GPU performance on specific deep learning tasks, this paper focuses on evaluating GPU performance at the operator level. We statistically analyze the information of operators on 12 representative deep learning models from six prominent AI tasks and provide an operator dataset to show the different importance of various types of operators in different networks. An operator-level benchmark, OpBench, is proposed on the basis of this dataset, allowing users to choose from a given range of models and set the input sizes according to their demands. This benchmark offers a detailed operator-level performance report for AI and hardware developers. We also evaluate four GPU models on OpBench and find that their performances differ on various types of operators and are not fully consistent with the performance metric FLOPS(floating point operations per second).

关键词： deep learning operator benchmark model benchmark GPU performance analysis deep neural networks