检索结果-南通市图书馆

Ensemble successor representations for task generalization in offline-to-online reinforcement learning

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7期67卷 240-255页

作者： Changhong wang xudong YU Chenjia BAI Qiaosheng ZHANG Zhen wang Space Control and Inertial Technology Research Center Harbin Institute of Technology Shanghai Artificial Intelligence Laboratory Shenzhen Research Institute of Northwestern Polytechnical University School of Cybersecurity Northwestern Polytechnical University

In reinforcement learning(RL), training a policy from scratch with online experiences can be inefficient because of the difficulties in exploration. Recently, offline RL provides a promising solution by giving an initialized offline policy, which can be refined through online interactions. However, existing approaches primarily perform offline and online learning in the same task, without considering the task generalization problem in offline-to-online adaptation. In real-world applications, it is common that we only have an offline dataset from a specific task while aiming for fast online-adaptation for several tasks. To address this problem, our work builds upon the investigation of successor representations for task generalization in online RL and extends the framework to incorporate offline-to-online learning. We demonstrate that the conventional paradigm using successor features cannot effectively utilize offline data and improve the performance for the new task by online fine-tuning. To mitigate this, we introduce a novel methodology that leverages offline data to acquire an ensemble of successor representations and subsequently constructs ensemble Q functions. This approach enables robust representation learning from datasets with different coverage and facilitates fast adaption of Q functions towards new tasks during the online fine-tuning phase.Extensive empirical evaluations provide compelling evidence showcasing the superior performance of our method in generalizing to diverse or even unseen tasks.

关键词： offline reinforcement learning online fine-tuning task generalization successor representations ensembles

Pore Formation Mechanism in W-C Hard Coatings Using Directed Energy Deposition on Tungsten Alloy

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Acta Metallurgica Sinica(English Letters) 2024年第1期37卷 89-101页

作者： Xinrui Zhang Weijie Fu Chen wang Zhenglong Lei Haoran Sun xudong Li State Key Laboratory of Advanced Welding and Joining Harbin Institute of TechnologyHarbin150001China

Porosity is a common phenomenon and can significantly hinder the quality of the coating.Here,the pore formation mechanism and the characteristics of the single tracks of the W-C coating using directed energy deposition(DED)are systematically investigated.The forming quality of the tracks,the distribution of the pores,and the elemental distribution near the pores are analyzed by the observations of the cross-sections of the tracks.The temperature field of the melt pool is discussed comprehensively to reveal the pore formation mechanism.The results confirm that Ni and Co evaporated during the DED process due to the high temperature of the melt pool.Pores were continuously produced adjacent to the fusion line when the melt pool was about to solidify since the temperature at the solidification front was higher than the boiling point of Ni.The vaporization area at the fusion line was proposed,where Ni could also evaporate at the time the melt pool started to solidify.The relationship between the solidification rate,the size of the vaporization area and the DED parameters(laser power and scanning speed)was established to discuss the causes of severe pores above the fusion line.This work contains a practical guide to reduce or eliminate the porosity in the coating preparation process on the surface of the tungsten alloy.

关键词： Directed energy deposition(DED)process Tungsten alloy W-C coating Pores Binder phase

Efficient stochastic parallel gradient descent training for on-chip optical processor

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Opto-Electronic Advances 2024年第4期7卷 5-15页

作者： Yuanjian Wan xudong Liu Guangze Wu Min Yang Guofeng Yan Yu Zhang Jian wang Wuhan National Laboratory for Optoelectronics and School of Optical and Electronic Information Huazhong University of Science and TechnologyWuhan 430074China Optics Valley Laboratory Wuhan 430074China

In recent years,space-division multiplexing(SDM)technology,which involves transmitting data information on multiple parallel channels for efficient capacity scaling,has been widely used in fiber and free-space optical communication sys-tems.To enable flexible data management and cope with the mixing between different channels,the integrated reconfig-urable optical processor is used for optical switching and mitigating the channel crosstalk.However,efficient online train-ing becomes intricate and challenging,particularly when dealing with a significant number of channels.Here we use the stochastic parallel gradient descent(SPGD)algorithm to configure the integrated optical processor,which has less com-putation than the traditional gradient descent(GD)algorithm.We design and fabricate a 6×6 on-chip optical processor on silicon platform to implement optical switching and descrambling assisted by the online training with the SPDG algorithm.Moreover,we apply the on-chip processor configured by the SPGD algorithm to optical communications for optical switching and efficiently mitigating the channel crosstalk in SDM systems.In comparison with the traditional GD al-gorithm,it is found that the SPGD algorithm features better performance especially when the scale of matrix is large,which means it has the potential to optimize large-scale optical matrix computation acceleration chips.

关键词： optical communications optical signal processing channel descrambling optical neural network chip silicon photonics

Atherosis-associated lnc_000048 activates PKR to enhance STAT1-mediated polarization of THP-1 macrophages to M1 phenotype

在线全文

学校读者我要写书评

暂无评论

Neural Regeneration Research 2024年第11期19卷 2488-2498页

作者： Yuanyuan Ding Yu Sun Hongyan wang Hongqin Zhao Ruihua Yin Meng Zhang xudong Pan Xiaoyan Zhu Department of Neurology The Affiliated Hospital of Qingdao UniversityQingdaoShandong ProvinceChina Qingdao Cadre Health Care Service Center The Affiliated Hospital of Qingdao UniversityQingdaoShandong ProvinceChina Department of Critical Care Medicine The Affiliated Hospital of Qingdao UniversityQingdaoShandong ProvinceChina

Our previous study has demonstrated that lnc_000048 is upregulated in large-artery atherosclerotic stroke and promotes atherosclerosis in ApoE^(-/-)mice.However,little is known about the role of lnc_000048 in classically activated macrophage(M1)polarization.In this study,we established THP-1-derived testing state macrophages(M0),M1 macrophages,and alternately activated macrophages(M2).Real-time fluorescence quantitative PCR was used to verify the expression of marker genes and the expression of lnc_000048 in macrophages.Flow cytometry was used to detect phenotypic proteins(CD11b,CD38,CD80).We generated cell lines with lentivirus-mediated upregulation or downregulation of lnc_000048.Flow cytometry,western blot,and real-time fluorescence quantitative PCR results showed that down-regulation of lnc_000048 reduced M1 macrophage polarization and the inflammation response,while over-expression of lnc_000048 led to the opposite effect.Western blot results indicated that lnc_000048 enhanced the activation of the STAT1 pathway and mediated the M1 macrophage polarization.Moreover,catRAPID prediction,RNA-pull down,and mass spectrometry were used to identify and screen the protein kinase RNA-activated(PKR),then catRAPID and RPIseq were used to predict the binding ability of lnc_000048 to PKR.Immunofluorescence(IF)-RNA fluorescence in situ hybridization(FISH)double labeling was performed to verify the subcellular colocalization of lnc_000048 and PKR in the cytoplasm of M1 macrophage.We speculate that lnc_000048 may form stem-loop structure-specific binding and activate PKR by inducing its phosphorylation,leading to activation of STAT1 phosphorylation and thereby enhancing STAT1 pathway-mediated polarization of THP-1 macrophages to M1 and inflammatory factor expression.Taken together,these results reveal that the lnc_000048/PKR/STAT1 axis plays a crucial role in the polarization of M1 macrophages and may be a novel therapeutic target for atherosclerosis alleviation in stroke.

关键词： atherosclerosis inflammation lnc_000048 lncRNA macrophage polarization protein kinase RNA-activated(PKR) STAT1

Effect of nitrile butadiene rubber hardness on the sealing characteristics of hydraulic O-ring rod seals

在线全文

学校读者我要写书评

暂无评论

Journal of Zhejiang University-Science A(Applied Physics & Engineering) 2024年第1期25卷 63-78页

作者： Xiaoxuan LI Bingqing wang xudong PENG Yuntang LI Xiaolu LI Yuan CHEN Jie JIN College of Quality and Safety Engineering China Jiliang UniversityHangzhou310018China College of Mechanical and Electrical Engineering China Jiliang UniversityHangzhou310018China College of Mechanical Engineering Zhejiang University of TechnologyHangzhou310023China

The nitrile butadiene rubber(NBR)hardness effect on the sealing characteristics of hydraulic O-ring rod seals is analyzed based on a mixed lubrication elastohydrodynamic model.Parameterized studies are conducted to reveal the mechanism of the influence of rubber hardness on the static and dynamic behavior of seals.The optimized selections of rubber hardness are then investigated under different conditions.Results show that the low hardness seal is prone to stress concentration due to the extrusion effect under high pressure conditions;it is also more prone to leaking.A high hardness seal can better prevent leakage by reducing film thickness but it will cause large frictional power loss and increase the probability of wear failure.The choice of low hardness is recommended to reduce friction with the premise that leakage requirements are met.

关键词： Nitrile butadiene rubber(NBR)hardness Sealing characteristics Optimized selection O-ring seal

RB-DEM Modeling and Simulation of Non-Persisting Rough Open Joints Based on the IFS-Enhanced Method

在线全文

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2024年第4期139卷 337-359页

作者： Hangtian Song xudong Chen Chun Zhu Qian Yin Wei wang Qingxiang Meng Key Laboratory of Ministry of Education for Geomechanics and Embankment Engineering Hohai UniversityNanjing210098China Research Institute of Geotechnical Engineering Hohai UniversityNanjing210098China School of Earth Sciences and Engineering Hohai UniversityNanjing210098China State Key Laboratory for Geomechanics and Deep Underground Engineering China University of Mining and TechnologyXuzhou221116China

When the geological environment of rock masses is disturbed,numerous non-persisting open joints can appear within it.It is crucial to investigate the effect of open joints on the mechanical properties of rock mass.However,it has been challenging to generate realistic open joints in traditional experimental tests and numerical simulations.This paper presents a novel solution to solve the problem.By utilizing the stochastic distribution of joints and an enhanced-fractal interpolation system(IFS)method,rough curves with any orientation can be generated.The Douglas-Peucker algorithm is then applied to simplify these curves by removing unnecessary points while preserving their fundamental shape.Subsequently,open joints are created by connecting points that move to both sides of rough curves based on the aperture distribution.Mesh modeling is performed to construct the final mesh model.Finally,the RB-DEM method is applied to transform the mesh model into a discrete element model containing geometric information about these open joints.Furthermore,this study explores the impacts of rough open joint orientation,aperture,and number on rock fracture mechanics.This method provides a realistic and effective approach for modeling and simulating these non-persisting open joints.

关键词： Non-persisting rough open joints stochastic distribution of joints enhanced-IFS method RB-DEM

Insight into the experiment and extraction mechanism for separating carbazole from anthracene oil with quaternary ammonium-based deep eutectic solvents

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Chemical Engineering 2024年第1期65卷 188-199页

作者： xudong Zhang Yanhua Liu Jun Shen Yugao wang Gang Liu Yanxia Niu Qingtao Sheng College of Chemical Engineering and Technology Taiyuan University of TechnologyTaiyuan 030024China

Carbazole is an irreplaceable basic organic chemical raw material and intermediate in industry.The separation of carbazole from anthracene oil by environmental benign solvents is important but still a challenge in chemical engineering.Deep eutectic solvents (DESs) as a sustainable green separation solvent have been proposed for the separation of carbazole from model anthracene oil.In this research,three quaternary ammonium-based DESs were prepared using ethylene glycol (EG) as hydrogen bond donor and tetrabutylammonium chloride (TBAC),tetrabutylammonium bromide or choline chloride as hydrogen bond acceptors.To explore their extraction performance of carbazole,the conductor-like screening model for real solvents (COSMO-RS) model was used to predict the activity coefficient at infinite dilution (γ^(∞)) of carbazole in DESs,and the result indicated TBAC:EG (1:2) had the stronger extraction ability for carbazole due to the higher capacity at infinite dilution (C^(∞)) value.Then,the separation performance of these three DESs was evaluated by experiments,and the experimental results were in good agreement with the COSMO-RS prediction results.The TBAC:EG (1:2) was determined as the most promising solvent.Additionally,the extraction conditions of TBAC:EG (1:2) were optimized,and the extraction efficiency,distribution coefficient and selectivity of carbazole could reach up to 85.74%,30.18 and 66.10%,respectively.Moreover,the TBAC:EG (1:2) could be recycled by using environmentally friendly water as antisolvent.In addition,the separation performance of TBAC:EG (1:2) was also evaluated by real crude anthracene,the carbazole was obtained with purity and yield of 85.32%,60.27%,respectively.Lastly,the extraction mechanism was elucidated byσ-profiles and interaction energy analysis.Theoretical calculation results showed that the main driving force for the extraction process was the hydrogen bonding ((N–H...Cl) and van der Waals interactions (C–H...O and C–H...π),which corresponding to the blue and green isosurfaces in IGMH anal

关键词： Carbazole Model anthracene oil Deep eutectic solvents COSMO-RS Extraction mechanism

维普期刊数据库博看期刊

Insights and reviews on battery lifetime prediction from research to practice

在线全文

学校读者我要写书评

暂无评论

Journal of Energy Chemistry 2024年第7期94卷 716-739页

作者： xudong Qu Dapai Shi Jingyuan Zhao Manh-Kien Tran Zhenghong wang Michael Fowler Yubo Lian Andrew FBurke Hubei Longzhong Laboratory Hubei University of Arts and ScienceXiangyang 441000HubeiChina Institute of Transportation Studies University of California DavisDavisCA 95616USA Department of Chemical Engineering University of WaterlooWaterlooOntario N2L 3G1Canada BYD Automotive Engineering Research Institute Shenzhen 518118GuangdongChina

The rising demand for energy storage solutions,especially in the electric vehicle and renewable energy sectors,highlights the importance of accurately predicting battery health to enhance their longevity and reliability.This article comprehensively examines various methods used to forecast battery health,including physics-based models,empirical models,and equivalent circuit models,among others.It delves into the promise of data-driven prognostics,utilizing both conventional machine learning and cuttingedge deep neural network techniques.The advantages and limitations of hybrid models are thoroughly analyzed,with a focus on the benefits of integrating diverse data sources to improve prognostic precision.Through practical case studies,the article showcases the effectiveness and flexibility of these approaches.It also critically addresses the challenges encountered in applying battery health prognostics in realworld scenarios,such as issues of scalability,complexity,and data anomalies.Despite these challenges,the article underscores the emerging opportunities brought about by recent technological,academic,and research advancements.These include the development of digital twin models for batteries,the use of data-centric Al and standardized benchmarking,the potential integration of blockchain technology for enhanced data security and transparency,and the synergy between edge and cloud computing to boost data analysis and processing.The primary goal of this article is to enrich the understanding of current battery health prognostic techniques and to inspire further research aimed at overcoming existing hurdles and tapping into new opportunities.It concludes with a visionary perspective on future research directions and potential developments in this evolving field,encouraging both researchers and practitioners to explore innovative solutions.

关键词： Batteries Lifetime Health Machine learning Deep learning Field Real world

Effect of Blood Circulation in Veins on Resonance Suppression of the Dragonfly Wing Constructed by Numerical Method

在线全文

学校读者我要写书评

暂无评论

Journal of Bionic Engineering 2024年第2期21卷 877-891页

作者： Lijun Zhang Xu Zhang Kaifei wang Zhenwei Gan Shibo Liu Xiao Liu Zhengjun Jing xudong Cui Jiahui Lu Jing Liu College of Mechanical and Electronic Engineering China University of PetroleumQingdao266580China

To reveal the resonance suppression mechanism of the blood circulation in dragonfly wings,a numerical modeling method of dragonfly wings based on Voronoi diagrams is proposed,and the changes in mass,aerodynamic damping,and natural frequencies caused by blood circulation in veins are investigated.The equivalent mass of blood,boundary conditions,and aerodynamic damping are calculated theoretically.Modal analysis and harmonic response analysis of wing models with different blood circulation paths are performed using the finite-element method(FEM).The vibration reduction ratioδis introduced to compare the damping efficiency of different mass regions.Finally,a natural frequency testing device is constructed to measure the natural frequencies of dragonfly wings.The results indicate that the shape,mass,and natural frequencies of the dragonfly wing model constructed by numerical method agree well with reality.The mass distribution on the wing can be altered by blood circulation,thereby adjusting the natural frequencies and achieving resonance suppression.The highestδof 1.013 is observed in the C region when blood circulates solely in secondary veins,but it is still lower than theδof 1.017 when blood circulates in complete veins.The aerodynamic damping ratio(1.19–1.79%)should not be neglected in the vibration analysis of the beating wing.

关键词： Dragonfly wings Blood circulation Resonance suppression Voronoi diagram FEM