基于数据场联合PRI变换与聚类的雷达信号分选

基于脉冲描述字进行雷达信号分选时,传统聚类算法需要预先人工设定聚类中心和聚类数目。针对该问题,提出一种基于数据场理论联合脉冲重复间隔(pulse repetition interval,PRI)变换与聚类的雷达信号分选新方法。首先,依据数据场理论,基于势值大小实现干扰点剔除,而后利用PRI变换算法进行PRI估计,依据PRI估计值将归一化脉冲描述字数据预分类,进而以各类数据集中心间的欧氏距离小于辐射因子为准则进行类别合并,自动得到初始聚类中心和聚类数目,最后通过改进K-Means算法完成聚类分选。仿真实验表明:所提方法能够应对存在频率捷变,重频参差、抖动、参数交叠、局部脉冲丢失的复杂信号环境,分选正确率明显提升。     

基于模糊证据推理的突发事件风险分析模型

突发事件发生、发展和演化过程中的风险属性来源复杂,且相关信息具有不确定、不完备的特点.为准确分析这一特点下的突发事件运行风险问题,引入模糊证据推理构建分析模型.首先根据突发事件的结构化描述框架和故障模式影响危害分析(FMECA)方法构建突发事件风险属性识别框架,获取信息作为风险分析证据并对其进行规范化处理;然后以模糊证据推理构建突发事件风险分析的模型,通过计算,给出事件风险程度的合理判断;最后选取CRH某型动车组行车安全事件进行实例分析.分析结果表明,在分析突发事件风险时,针对信息不确定、不完备特点构建的模糊证据推理模型具有可行性与有效性,同时该模型能够为突发事件风险分析和风险应对决策的制定提供辅助参考.     

横向合成背景的面向对象概念格的生成

针对多形式背景合成的概念格生成,提出了一种多背景横向合成的面向对象概念格的生成方法。首先利用各个子形式背景中的每个概念的外延通过相关运算得到合成背景的面向对象概念格的外延集,然后将得到的外延集通过近似算子运算得到合成背景的内涵,并进一步得到合成背景的概念格。用理论证明该方法的有效性和完备性,并通过实验表明该方法时间效率较好。     

基于神经网络的图像描述方法研究

自动识别和描述图像的内容是人工智能中一个重要的研究方向,它涉及计算机视觉和自然语言处理技术。针对这一难题,提出了一种由深层神经网络模型生成自然语言句子来描述图像内容的方法。该方法提出的模型由卷积神经网络(Convolution Neural Network,CNN)和循环神经网络(Recurrent Neural Network,RNN)组成,其中,CNN用来提取输入图像的特征生成固定长度的特征向量,该特征向量初始化RNN来生成句子。在MSCOCO图像描述数据集上的实验结果表明了该模型所生成句子的语法准确性和语义准确性,且优于先前的基线模型。     

Taxonomy of the genus Tantunema Siddiqi, 1982 (Dorylaimida: Tylencholaimidae) with description of one new and two known species

One new and two known species of the genus Tantunema Siddiqi, 1982 are described and illustrated. Tantunema indicum sp. nov. is characterized by having a small body (length 0.32–0.45 mm); lip region cap-like, set off by slight constriction; odontostyle 6–7 µm and odontophore 8–10 µm long; pharyngeal expansion gradual, expanded part occupying about 35–41% of total neck length, lumen of posterior expanded part widened and thickened; female genital system mono-opisthodelphic, anterior genital branch absent; tail short, hemispheroid, 0.8–1.1 anal body diameter long. Tantunema pakistanense (Timm, 1964) comb. nov. (=Tylencholaimus pakistanensis Timm, 1964) and Tantunema intermedium (Peña-Santiago and Coomans, 1996) Andrássy, 2009 are redescribed. A diagnostic compendium and key to species of the genus is also provided.     

Larval stages of 17 rare and poorly known British hoverflies (Diptera: Syrphidae)

Third stage larvae of three species of phytophagous (Portevina maculata, Cheilosia albitarsis and C. antiqua), thirteen species of saprophagous (Callicera aurata, Brachyopa bicolor, B. insensilis, B. pilosa, Hammerschmidtia ferruginea, Myolepta potens, Brachypalpus laphriformis, Caliprobola speciosa, Criorhina berberina, C. floccosa, C. ranunculi, Pocota personata and Xylota tarda) and one possibly predatory species (Microdon devius) of Syrphidae are described. Most of the saprophagous species breed in dead wood and many are European rarities. Nine of the 17 species described are included in the British Red Data Books: 2. Insects (Shirt, 1987). For the larva of each species, recognition features and breeding sites are described.     

New species of Hygiella Mesnil (Diptera: Tachinidae), parasitoids of leaf insects (Phasmatodea: Phylliidae)

The Oriental genus Hygiella Mesnil is revised: three species are described as new, Hygiella angustifrons sp. nov., Hygiella luteipes sp. nov. and Hygiella proclinata sp. nov., and two known species are re-described, Hygiella nigripes Mesnil and Hygiella pygidialis Mesnil. A key to these five species is provided. The host of H. proclinata, Phyllium siccifolium (Linnaeus) (Phasmatodea: Phylliidae), is recorded for the first time as a host of the Tachinidae. The systematic position of Hygiella is discussed and the genus is placed in the tribe Acemyini (Exoristinae) based on the structure of the male terminalia.     

STS视域下创新型企业内涵分析与特征描述

基于STS(科学技术与社会)视角,系统分析了创新型企业的内涵,并对其特征做了初步的描述,以期有益于创新型企业(试点)工作的开展。     

基于Petri网的电梯系统规格说明

为克服非形式化技术描述系统规格说明带来的二义性,采用了一种形式化技术——Petri网来描述电梯系统的规格说明。Petri网技术是对离散并行系统的数学表示,适合于描述并发的计算机系统模型,可以正确的描述电梯系统。     

基于高维几何特征的孤立点检测算法

针对孤立点检测算法不能较好地模拟人工检测过程、未充分考虑待测数据周围样本分布的问题,提出了一种孤立点检测算法DD-SVDD.该算法综合考虑待测样本与目标样本之间的距离,以及待测样本所在区域样本的分布信息,结合距离和平均密度来确定高维特征空间中决策边界附近待测数据的类别.在训练阶段,考虑了决策边界附近目标训练样本的分布,...