Overflowing phenomenon during ultrasonic treatment in Al-Si alloys

At the late stage of solidification with ultrasonic treatment （UST） in Al-Si alloys, a part of semisolid overflows and climbs along the probe. The interesting phenomenon and its influence on the solidification microstructure were investigated in order to better study the mechanism of UST. It is considered that the overflowing phenomenon occurs due to the changes of vibration and flow in the remaining semisolid. Because the overflowed portion comes from the region with intense UST effect and vibrates with the probe during solidification, great modification of primary and euteetic Si （about 10 pm in length） and refinement of primary a（Al） （about 70 μm in size） are observed in this portion.

Pre-emptive Channel Borrowing and Traffic Overflowing Channel Allocation Scheme for Multimedia Overlay Networks

This paper proposes a channel allocation scheme for multimedia wireless networks,in which a twolayer macro-cell or micro-cell architecture is considered.Macro-cells are used to access high-mobility services;while micro-cells,which are overlaid by the macro-cells,are used to cater low-mobility services.To analyze the scheme,a multidimensional Markov traffic model is firstly developed,in which traffic characteristic of two special periods of time is considered.And then,a pre-emptive channel-borrowing scheme combined with trafficoverflowing strategy for multimedia(voice,video or data) networks is proposed,in which handoff requests can not only borrow channels from adjacent homogenous cells,but also be overflowed to heterogeneous cells.Priority strategies are also dedicated to high-mobility services for they can pre-empt channels being used by low-mobility services in macro-cells.To meet the high quality of service(QoS) requirements of video services and increase the channel utilization ratio,video services can be transformed between real-time services and non-real-time services as necessary.Simulation results show that our schemes can decrease the blocking probabilities and improve the channel utilization.

Chinese lanterns overflowing with lights, colors in Europe

In the winter of 2017,southern France was hit by the most severe freeze in a decade,and yet the southern city of Gaiaque,with a population of just 10,000,had 250,000 visitors in town.What attracted them to brave the cold and rain to visit the small city?A Chinese lantern festival,the first kind of it in France.The lantern festival,jointly

一种结合代码片段和混合主题模型的软件数据聚类方法

使用主题模型进行文档聚类是众多文本挖掘任务中一种常见的做法。许多研究针对软件问答网站的数据,利用主题模型进行聚类来分析不同领域在社区的发展情况。然而,这些软件相关数据往往包含代码片段且文本长度分布不均,使用传统单一的主题模型对文本数据建模,易得到不稳定的聚类结果。文中提出了一种结合代码片段和混合主题模型的聚类方法,并使用Stack Overflow作为数据源,构造了在该平台上被提问数量排名前60的Python第三方库数据集,经过建模,该数据集最终划分为以下6个不同的领域:网络安全、数据分析、人工智能、文本处理、软件开发和系统终端。实验结果表明,在自动评估和人工评估的指标上,使用代码片段结合文本进行主题建模,在聚类结果划分的质量上表现良好,而联合多个模型进行实验,一定程度上提高了聚类结果的稳定性和准确性。

Effects of retained dry material on the impact,overflow and landing dynamics

During long-term operation,the performance of obstacles would be changed due to the material accumulating upslope the obstacle.However,the effects of retained material on impact,overflow and landing dynamics of granular flow have not yet been elucidated.To address this gap,physical flume tests and discrete element simulations are conducted considering a range of normalized deposition height h0/H from 0 to 1,where h0 and H represent the deposition height and obstacle height,respectively.An analytical model is modified to evaluate the flow velocity and flow depth after interacting with the retained materials,which further serve to calculate the peak impact force on the obstacle.Notably,the computed impact forces successfully predict the experimental results when a≥25°.In addition,the results indicate that a higher h0/H leads to a lower dynamic impact force,a greater landing distance L,and a larger landing coefficient Cr,where Cr is the ratio of slope-parallel component of landing velocity to flow velocity just before landing.Compared to the existing overflow model,the measured landing distance L is underestimated by up to 30%,and therefore it is insufficient for obstacle design when there is retained material.Moreover,the recommended Cr in current design practice is found to be nonconservative for estimating the landing velocity of geophysical flow.This study provides insightful scientific basis for designing obstacles with deposition.

Optimized parameters of downhole all-metal PDM based on genetic algorithm

Currently,deep drilling operates under extreme conditions of high temperature and high pressure,demanding more from subterranean power motors.The all-metal positive displacement motor,known for its robust performance,is a critical choice for such drilling.The dimensions of the PDM are crucial for its performance output.To enhance this,optimization of the motor's profile using a genetic algorithm has been undertaken.The design process begins with the computation of the initial stator and rotor curves based on the equations for a screw cycloid.These curves are then refined using the least squares method for a precise fit.Following this,the PDM's mathematical model is optimized,and motor friction is assessed.The genetic algorithm process involves encoding variations and managing crossovers to optimize objective functions,including the isometric radius coefficient,eccentricity distance parameter,overflow area,and maximum slip speed.This optimization yields the ideal profile parameters that enhance the motor's output.Comparative analyses of the initial and optimized output characteristics were conducted,focusing on the effects of the isometric radius coefficient and overflow area on the motor's performance.Results indicate that the optimized motor's overflow area increased by 6.9%,while its rotational speed reduced by 6.58%.The torque,as tested by Infocus,saw substantial improvements of38.8%.This optimization provides a theoretical foundation for improving the output characteristics of allmetal PDMs and supports the ongoing development and research of PDM technology.

Stack Overflow上机器学习相关问题的大规模实证研究

为了调查机器学习相关主题分布和发展趋势,从在线问答网站Stack Overflow上,利用过滤标签,从4 178多万帖子中提取出60 028个与机器学习相关的问题帖.通过分析问题帖,统计各个机器学习平台的讨论量,发现Scikit-learn、TensorFlow、Keras是前3位频繁被讨论的机器学习平台,占总讨论量的58%.为了进一步分析机器学习相关讨论主题,进行潜在狄利克雷分布(LDA)主题模型训练,提出自适应LDA中的主题数渐进搜索方法,采用主题一致性系数评估输出结果,获得主题最佳数量,从而发现9个讨论主题,分属3个类别:代码相关、模型相关、理论相关.基于主题中问题帖的浏览数、评论数,分析不同主题的流行度和回答困难程度.

台山核电厂淡水水源工程水库溢流坝消能试验研究

以台山核电厂淡水水源工程为例,通过水力模型试验研究,推荐溢流坝采用宽尾墩+坝面削角阶梯+底流消力池的联合消能方案。试验结果表明,该方案消能效果较显著、工程量小,优化了溢流坝工程布置和体型。

Experimental study on inner slope failure mechanism of seawall by coupling effect of storm surge and wave

In the context of global climate change,the impact of group-occurring ocean dynamic disasters on China's offshore areas is becoming more and more intense.The study of the effect of existing ocean dynamic disasters on offshore hazard-bearing bodies mostly focuses on the effect of single disaster-causing factors,and it is still insufficient to study storm surge and dynamic wave coupling&reinforcement effects as well as the process of the dynamic response of such hazard-bearing bodies as seawalls.This study firstly realized the synchronous process of water level and wave through continuous tide generation and wave generation by the wave maker and tide generating device,so as to realize the dynamic coupling simulation of storm surge and wave in the laboratory.Then the physical model test of the typical seawall section was carried out under the dynamic coupling of storm surge and wave as well as at a conventional fixed water level respectively.In the process of test wave overtopping discharge and the damage process of the levee crown and backwall of seawalls were observed and compared,and their damage mechanism was also studied.

Numerical Study of Storm Surge Inundation in the Southwestern Hangzhou Bay Region During Typhoon Chan-Hom in 2015

Storm surge inundation is a major concern in marine hazard risk assessment during extreme weather conditions.In this study,a high-resolution coupled model(the ADVanced CIRCulation model+the Simulating WAves Nearshore model)was used to investigate the storm surge inundation in the southwestern Hangzhou Bay region during Typhoon Chan-hom in 2015.The simulated hydrodynamic processes(sea surface wave and storm tide)were validated with measured data from wave buoys and tide gauges,indicating that the overall performance of the model was satisfactory.The storm surge inundation in the coastal area was simulated for several idealized control experiments,including different wave effects(wave-enhanced wind stress,wave-enhanced bottom stress,and wave radiation stress).Dike overflowing cases with different dike heights and dike breaking cases with different dike breach lengths were considered in the simulation.The results highlight the necessity of incorporating wave effects in the accurate simulation of storm surge inundation.Dike height significantly influences the magnitude and phase of the maximum inundation area in the dike overflowing cases,and dike breach length is an important factor impacting the magnitude of the maximum inundation area in the dike breaking cases.This study may serve as a useful reference for accurate coastal inundation simulation and risk assessment.

An improvement to the simulation of phase-change heat-transfer during soil freezing and thawing

A new variable time step method,which is called the backwards calculating time step method,is presented in this paper.It allows numerical simulation of soil freezing and thawing while avoiding "phase change missing and overflowing".A sensitive heat capacity model is introduced through which the calculation errors are analyzed.Then the equation using the self-adjusted time step is presented and solved using finite differences.Through this equation,the time needed for a space cell to reach the phase change point temperature is calculated.Using this time allows the time step to be adjusted so that errors caused by "phase change missing and overflowing" are successfully eliminated.Above all,the obvious features of this method are an accelerated rate for adjusting the time step and simplifing the computations.An actual example proves that this method can accurately calculate the temperature fields during soil freezing and thawing.It is an improvement over traditional methods and can be widely used on complicated multi-dimensional phase change problems.

基于数据挖掘的专业可信回答者个性化推荐——以Stack Overflow问答社区为例

针对问答社区中问题不能得到及时、有效解答的现状,文章以Stack Overflow问答社区为例,首先介绍了问答社区数据的采集与预处理情况;然后,通过挖掘学习者信息,得到专业可信回答者、高信誉回答者和徽章回答者三种潜在回答者;最后,实施了三种回答者推荐并对比了推荐性能。实验结果表明,与高信誉回答者推荐和徽章回答者推荐相比,考虑回答质量和专业相关性的专业可信回答者推荐具有更高的准确率和召回率,其推荐性能更优。实施基于数据挖掘的专业可信回答者个性化推荐,能有效缓解问答社区的信息过载问题,有助于建设更高效的网络学习社区环境。

基于问答语义匹配的知识社区新问题专家推荐方法

传统的知识社区专家推荐方法采用文本相似度匹配机理,并基于问题或专家描述来构建专家特征.这些方法没有利用问题与答案的语义匹配关系,因此难以充分挖掘专家回答问题的能力特征,影响推荐性能.提出一种基于综合历史和当前问答语义匹配的知识社区新问题的专家推荐方法(History-Now Semantics Expert RECommendation model,HNS-EREC).首先,采用反馈评价和负采样技术来处理数据集中的两类不平衡现象;其次,基于问答语义来提取专家回答问题能力特征;最后,提出一种基于问答语义匹配的History-Now联合专家推荐模型,该模型能够实现面向专家的历史问答和当前问答的语义联合学习.实验结果表明,相对于其他方法,本文所提出的HNS-EREC方法在新问题专家推荐方面具有显著的优势.

Numerical simulation of cup hydrodynamic deep drawing

The simulation of hydrodynamic deep drawing by means of FEM is an efficient method that can relieve experimental burden and find the optimum process parameters. Some problems such as mathematical description of cavity liquid flow pressure must be solved firstly. A math formula about hydrodynamic flow pressure that can be applied in general FEM software was proposed, and good results were gained. It was proved that the theoretical results keep coincident with experimental results.[

A New Channel Allocation Scheme for Hierarchical Wireless Networks

A channel allocation scheme for hierarchical wireless networks was proposed in terms of the connection-level quality of service. The channel allocation scheme was analyzed using both horizontal channel borrowing and vertical traffic overflowing. Pre-emptive priority strategies are used to classify real-time services and non-real-time services, real-time service is given higher priority for it is allowed to pre-empt channels used by non-real-time service. Some channel borrowing thresholds and acceptance ratios are used to avoid channel locking or dynamic power control, which can also be dynamically adjusted according to network load. Simulation results show that the proposed schemes can improve the system performance.

Method of integer overflow detection to avoid buffer overflow

A simplified integer overflow detection method based on path relaxation is described for avoiding buffer overflow triggered by integer overflow. When the integer overflow refers to the size of the buffer allocated dynamically, this kind of integer overflow is most likely to trigger buffer overflow. Based on this discovery, through lightly static program analysis, the solution traces the key variables referring to the size of a buffer allocated dynamically and it maintains the upper bound and lower bound of these variables. After the constraint information of these traced variables is inserted into the original program, this method tests the program with test cases through path relaxation, which means that it not only reports the errors revealed by the current runtime value of traced variables contained in the test case, but it also examines the errors possibly occurring under the same execution path with all the possible values of the traced variables. The effectiveness of this method is demonstrated in a case study. Compared with the traditional buffer overflow detection methods, this method reduces the burden of detection and improves efficiency.

Lossless Mapping from Semi-Structured Data to Structured Data

Most semi-structured data are of certain structure regularity. Having beenstored as structured data in relational database (RDB), they can be effectively managed by databasemanagement system (DBMS). Some semi-structured data are difficult to transform due to theirirregular structures. We design an efficient algorithm and data structure for ensuring losslesstransformation. We bring forward an approach of schema extraction through data mining, in whichdifferent kinds of elements are transformed respectively and lossless mapping from semi-structureddata to structured data can be achieved.

基于Stack Overflow的数据库相关主题分析

数据库管理系统虽是一种较为成熟的软件系统,但开发人员在应用数据库系统进行数据管理以及数据分析时还是会遇到各种问题,因此会在Stack Overflow之类的问答论坛上寻求解决方法。文中获取了Stack Overflow上94473条与数据库相关的问题,应用LDA主题模型将这些问题归为25个主题,结果显示开发者的问题可归为"表""SQL""SELECT"等主题。通过研究与数据库相关的不同主题的流行度和困难程度发现,"SQL"主题相关的问题较为流行。除此以外,文中还分别研究了3种不同的数据库,即MySQL,Oracle和MongoDB,分析了与不同数据库系统相关的问题的主题分布。文中的研究成果有助于了解数据库开发者所面临的挑战,从而为数据库系统版本更新、数据库课程教学内容的设置,甚至是数据库领域的研究问题提供参考。

Stack Overflow的缺陷代码特征分析与相似缺陷检测

目前在软件代码缺陷审查以及缺陷预测中,研究人员对源代码进行分析研究却忽略了代码的缺陷信息.本文通过对缺陷信息进行分析,发现缺陷信息对于相似缺陷的检测有着重要的参考价值.基于这一思想,本文分析软件缺陷社区Stack Overflow中关于缺陷代码的信息,提出一种基于缺陷代码特征分析的相似缺陷检测方法.该方法首先对缺陷报告进行LDA主题分析并将缺陷报告分类到不同的主题(类别)中,统计得到高频缺陷类别;其次对于高频缺陷类别的缺陷代码提取特征;最后根据缺陷代码特征构建相似缺陷检测模型.为了验证相似缺陷检测模型的有效性,针对数据操作缺陷数据构建诊断模型并对该模型进行实证,实验结果表明该方法对检测其他代码中相似缺陷有较好的效果.

基于CBOW-LDA主题模型的Stack Overflow编程网站热点主题发现研究

Stack Overflow是一个热门的国外编程问答网站,通过对该网站编程提问帖的问题文本进行文本语义挖掘,能获析用户关注的编程热点。由于研究对象所代表的短文本信息具有高维性及分布不均的特点,易导致主题获取不明晰。文中提出一种基于LDA(Latent Dirichlet Allocation)主题模型的CBOW-LDA建模方法,该方法对目标语料进行相似词聚类后再完成主题建模,能有效降低文本输入维度,使主题分布更明确。采集Stack Overflow网站上2010-2015年的问题帖数据集POST,并对其进行实验,同等主题数下采用文本建模中衡量模型性能的评价指标困惑度(Perplexity)来度量算法在不同数据集容量维度下的性能。结果表明,与现有的基于词频权重的词量化主题建模TFLDA方法相比,CBOW-LDA方法的困惑度更低,在实验语料下的困惑度降低约4.87%,证明了所提算法的性能更好。采用CBOW-LDA方法对Stack Overflow进行热点挖掘,同时使用TF-LDA方法进行对比实验,建立手工标注的标准评测集对两种方法获取的热门主题和热搜词汇进行查全率、查准率及F1值的判定,结果证实CBOW-LDA表现更佳,其热点挖掘效果较好。由实验结果可知,Java为该编程网站提问帖中最热门的主题,而C和Javascript则为该网站用户提问中被提及得最频繁的词汇。