AUTOMATIC DISPLACEMENT ANALYSIS OF PLANAR MECHANISMS WITH MATCHING BASIC KINEMATIC CHAINS METHOD

A simple, efficient, automatic and universal method to fulfill displacementanalysis of a great deal of mechanisms regenerated by Yan's mechanism creative theory has beendeveloped. For a regenerated mechanism, at first, the method identifies its type and structure andchanges it into a rigid structure by fixing the ground link and the input link. And then this rigidstructure is decomposed into a set of basic kinematic chains (BKCs). By matching the type of BKC,the displacement analysis equations can be set up, and all possible configurations, in whichpositions of all movable links are considered, can be given out.

Failure mechanism and influencing factors of cement sheath integrity under alternating pressure

The failure of cement sheath integrity can be easily caused by alternating pressure during large-scale multistage hydraulic fracturing in shale-gas well.An elastic-plastic mechanical model of casing-cement sheath-formation(CSF)system under alternating pressure is established based on the Mohr-Coulomb criterion and thick-walled cylinder theory,and it has been solved by MATLAB programming combining global optimization algorithm with Global Search.The failure mechanism of cement sheath integrity is investigated,by which it can be seen that the formation of interface debonding is mainly related to the plastic strain accumulation,and there is a risk of interface debonding under alternating pressure,once the cement sheath enters plasticity whether in shallow or deep well sections.The matching relationship between the mechanical parameters(elastic modulus and Poisson's ratio)of cement sheath and its integrity failure under alternating pressure in whole well sections is studied,by which it has been found there is a“critical range”in the Poisson's ratio of cement sheath.When the Poisson's ratio is below the“critical range”,there is a positive correlation between the yield internal pressure of cement sheath(SYP)and its elastic modulus.However,when the Poisson's ratio is above the“critical range”,there is a negative correlation.The elastic modulus of cement sheath is closely related to its Poisson's ratio,and restricts each other.Scientific and reasonable matching between mechanical parameters of cement sheath and CSF system under different working conditions can not only reduce the cost,but also protect the cement sheath integrity.

Reliability analysis of the velocity matching of coal cutting and caving in fully mechanized top-coal caving face

The matching relationship between coal cutting and caving in fully mechanized top-coal caving face is analyzed in detail from the angle of reliability. The coupling equation of reliability is established correspondingly, and the mathematical equation of the coefficient of velocity matching of coal cutting and caving is obtained, which meets a certain reliability demand for making the working procedure of coal caving not influence coal cutting of coal-cutter. The results show that the relationship between the coefficient of the velocity matching and the reliability of coal cutting and caving system is linear on the whole when R <0.9. It is pointed out that different numerical value should be selected for different coal face according to different demand for reliability.

Mechanisms and experimental research of ion-matched waterflooding to enhance oil recovery in carbonate reservoirs: A case of Cretaceous limestone reservoirs in Halfaya Oilfield, Middle East

Based on systematically summarizing the achievements of previous ion-matched waterflooding researches,the diversity and synergy of oil recovery enhancement mechanisms and the interaction between mechanisms are examined according to two classification standards,and the influence of behaviors of different ions on different mechanisms and oil displacement efficiency are investigated.Ionic strength is proposed to characterize the behavior differences of univalent and divalent ions,the relationships between ionic strength,effective concentration,and mechanisms are established to characterize the ion behavior behind various mechanisms,and evaluate the performance of ion-matched injection water.The mechanisms of enhancing oil recovery by ion-matched waterflooding include:(1)The ion-matched water can reduce the ion strength and match the ion composition of formation water,thereby reducing the difference between the effective concentration of univalent ions and divalent ions on the surface of carbonate rocks,and improving the effective concentration of potential determining ions(especially SO42-).(2)It can improve wettability,oil-water interface properties,pore structure and physical properties of the reservoir,and finally enable the establishment of a new ionic equilibrium conducive to waterflooding while breaking the original equilibrium.In this study,experiments such as relative permeability curve,interfacial tension,and core-flooding were carried out on carbonate core samples from the Cretaceous Mishrif Formation reservoirs in Halfaya Oilfield,Middle East,a method for injection water evaluation was established and the injection water suitable for these reservoirs was selected:6 times diluted seawater.Compared with ordinary seawater,oil displacement efficiency can be increased by more than 4.60%and compared with the optimum dilution of formation water,oil displacement efficiency can be increased by 3.14%.

Crystallography of Zr poisoning of Al-Ti-B grain refinement using edge-to-edge matching model

The mechanism of zirconium poisoning on the grain-refining efficiency of an Al-Ti-B based grain refiner was studied. The experiment was conducted by melting Al-5Ti-1B and Al-3Zr master alloys together. The edge-to-edge matching model was used to investigate and compare the orientation relationships between the binary intermetallic compounds present in the Al-Ti-B-Zr system. The results show that the poisoning effect probably results from the combination of Al3 Zr with Al3 Ti and the decreased amount of Ti solute, for Al3 Ti particles have good crystallographic relationships with Al3 Zr. Totally six orientation relationships may present between them, while they play vital roles in grain refinement. TiB2 particles appear to remain unchanged because of a bit large misfit. Only one orientation relationship may present between them to prevent Al3 Zr phase from forming on the surface of TiB2, though TiB2 is agglomerated. The theoretical calculation agrees well with the experimental results. The edge-to-edge matching model is proved to be a useful tool for discovering the orientation relationships between phases.

On Model of Translation Beginners' Mechanism of Mental Translation

Translation process is a process of information processing, and also a process of translators' thinking activity. During the process of translation or bilingual communication, translators or bilingual users employ mental translation. To expound how beginners communicate or translate in the bilingual environment, the model of mechanism of mental translation, i.e. match-making mechanism of mental translation, is theoretically constructed.

Numerical investigations on performance improvement mechanism of a high-power vertical centrifugal pump with special emphasis on hydraulic component matching

The purpose of this paper is to investigate the performance improvement mechanism of a high power vertical centrifugal pump by using numerical calculations.Therefore,a comparative study of energy losses and internal flow characteristics in the original and optimized models was carried out with special attention to the hydraulic component matching.The optimized model(model B)was obtained by optimizing the vaned diffuser and volute based on the original model(model A),mainly the diffuser inlet diameter,diffuser inlet vane angle,volute channel inlet width and volute throat area were changed.Firstly,the comparative results on performance and energy losses of two models showed that the efficiency and head of model B was significantly increased under design and part-load conditions.It is mainly due to the dramatic reduction of energy loss PL in the diffuser and volute.Then,the comparisons of PL and flow patterns in the vaned diffuser showed that the matching optimization between the model B impeller outlet flow angle and diffuser inlet vane angle resulted in a better flow pattern in both the circumferential and axial directions of the diffuser,which leads to the PL3 reduction.The meridian velocity Vm of model B was significantly increased at diffuser inlet regions and resulted in improvements of flow patterns at diffuser middle and outlet regions as well as pressure expansion capacity.Finally,the comparisons of PL and flow characteristics in the volute showed that the turbulence loss reduction in the model B volute was due to the flow pattern improvement at diffuser outlet regions which provided better flow conditions at volute inlet regions.The matching optimization between the diffuser and volute significantly reduced the turbulence loss in volute sections 1–4 and enhanced the pressure expansion capacity in sections 8–10.

基于注意力与密集重参数化的目标检测算法

针对目标检测任务中背景复杂、目标尺寸差异大等因素导致目标检测结果较差的问题,本文提出基于注意力和密集重参数化的目标检测算法。首先,基于CSP-DarkNet提出高效的特征提取网络,主要包括密集重参数化模块和CASA模块2个设计。前者利用密集连接保留浅层特征,又通过重参数化结构降低网络复杂度;后者CASA模块用于获取需要的目标信息。其次,特征融合在特征金字塔(FPN)和路径聚合网络(PAN)的基础上,引入内容感知特征重组(CARAFE)进行上采样,有效解决了邻近插值法等未能捕捉丰富语义信息的问题;提出更高效的C3-G模块,获取丰富的梯度信息,增强模型表达能力和感知能力;同时,引入深度可分离卷积提升运算效率。最后,检测输出采用在更大范围上跨领域正负样本匹配策略扩充正样本数量,提升检测效果。该算法在MS COCO和PASCAL VOC数据集上的mAP@0.5分别达到了57.5%和83.0%,充分说明了本文算法的先进性。

融合梯度先验与全局聚合的立体匹配算法

现有基于学习的立体匹配算法具有一定局限性,通常特征提取部分使用的常规卷积倾向于捕捉外观信息,忽略了结构信息的提取,并且由于卷积感受野的限制,这些算法在弱纹理、重复纹理等病态区域性能较差.为此本文提出了一种融合梯度信息及全局代价聚合的立体匹配网络模型(GFGANet).首先,设计了差分卷积残差组(CDC-ResBlock-Groups),显式编码结构信息提高特征表达,然后设计全局代价聚合模块(Trans-Hourglass),有效聚合代价体并捕获全局上下文,最后采用加窗视差回归降低了视差标签不平衡的影响,此外使用StereoDRNet的后处理进行视差细化.实验表明,所提方法在重复纹理,弱纹理区域表现优异,在SceneFlow中的端点误差(EPE)仅为0.645px,KITTI 2015上的误匹配率为2.04%,与基础模型PSMNet相比,性能显著提升.

数智时代科技成果供需对接机制

为探究我国数智时代科技成果供需对接影响因素和机制作用,梳理我国科技成果供需对接模式及演变历程,构建科技成果供需对接影响因素概念模型,运用扎根理论和文献分析方法归纳出供方、需方、平台、政策、资金和技术因素等影响因素,运用DEMATEL方法分析影响因素之间的影响度。基于分析结果,提出我国科技成果供需对接机制为从应需引供模式转变为掘需育供模式,从上下游关系转变为共同体关系,从局域平台转变为广域平台,从信息化转变为智慧化,从单点服务转变为全链条服务,从技术发展转变为多要素融合发展。

基于轨迹扰动和路网匹配的位置隐私保护算法

针对现有扰动机制未考虑位置点语义关系导致数据可用性较低的问题,提出一种基于差分隐私(DP)的轨迹位置隐私保护机制(DP-TLPM)。首先,DP-TLPM利用滑动窗口提取轨迹停留点生成模糊区域,再利用指数机制和拉普拉斯机制对该区域进行采样;其次,为了消除采样点中可能存在的无语义位置点,提出一种路网匹配算法,对轨迹分段并利用误差椭圆匹配(EEM)进行迭代匹配;最后,根据匹配后的位置点形成扰动轨迹,由用户端将扰动轨迹发送至服务器。实验以混淆质量和均方根误差(RMSE)为评价标准对该机制进行综合评测。与GeoInd算法相比,DP-TLPM的数据质量损失降低了24%,轨迹的混淆质量提高了52%,从隐私保护强度和数据质量两方面验证了该算法的有效性。

基于仿斑马鱼和仿鹰眼视觉的复杂背景下目标识别

针对反制无人机识别系统在公共场所内部复杂背景下的无人机识别问题,该文研究了一种基于仿斑马鱼模板匹配视觉识别和仿鹰眼视觉注意的目标识别方法,通过建立不同姿态的无人机模板数据库,采用仿鹰眼视觉搜索机制,结合尺度不变特征变换,将姿态模板图像与目标进行匹配,获得粗略的目标区域。然后计算模板姿态与目标姿态的Hausdorff距离,比较目标姿态相似性,获得最相似姿态。采用仿鹰眼视觉注意机制对遮挡图像进行处理,提高目标识别的显著性。实验结果表明,该方法能够在不同复杂背景下实现无人机的准确识别,与光谱残差的显著性目标识别方法相比,平均运行时间提高23.5%,与差异哈希算法相比,具有更高的结构相似性指数。

引入轻量级Transformer的自适应窗口立体匹配算法

现有端到端的立体匹配算法为了减轻显存消耗和计算量而预设固定视差范围,在匹配精度和运行效率上难以平衡。提出一种基于轻量化Transformer的自适应窗口立体匹配算法。利用具有线性复杂度的坐标注意力层对低分辨率特征图进行位置编码,减轻计算量并增强相似特征的辨别力;设计轻量化Transformer特征描述模块,转换上下文相关的特征,并引入可分离多头自注意力层对Transformer进行轻量化改进,降低Transformer的延迟性;用可微匹配层对特征进行匹配,设计自适应窗口匹配细化模块进行亚像素级的匹配细化,在提高匹配精度的同时减少显存消耗;经视差回归后生成无视差范围的视差图。在KITTI2015、KITTI2012和SceneFlow数据集上的对比实验表明,该算法比基于标准Transformer的STTR在匹配效率上快了近4.7倍,具有更快的运行速度和更友好的存储性能;比基于3D卷积的PSMNet误匹配率降低了18%,运行时间快了5倍,实现了更好的速度和精度的平衡。

基于特征增强和语义相关性匹配的图像文本检索方法

为实现图像文本检索中图像与文本的精确语义连接,提出一种基于特征增强和语义相关性匹配(FESCM)的图像文本检索方法。首先,通过特征增强表示模块,引入多头自注意力机制增强图像区域特征和文本单词特征,以减少冗余信息对图像区域和文本单词对齐的干扰;其次,通过语义相关性匹配模块,不仅利用局部匹配捕获局部显著对象之间的对应相关性,还把图像背景信息融入图像全局特征,利用全局匹配实现精确的全局语义相关性;最后,通过局部匹配分数和全局匹配分数获取图像和文本的最终匹配分数。实验结果表明,基于FESCM的图像文本检索方法在Flickr8k和Flickr30k基准数据集上的召回率总值比扩展的视觉语义嵌入方法分别提升了5.7和7.5个百分点,在MS-COCO数据集比双流层次相似度推理方法提升了3.7个百分点。因此该方法可以有效提高图像文本检索的准确度,实现图像与文本的语义连接。

高海拔铁路隧道斜井机械化配套与快速施工

为解决高海拔高寒地区斜井施工存在气压低、含氧量低、温差大、严寒干燥等问题,以某高海拔铁路隧道斜井工程为背景,结合围岩等级、断面大小、海拔高度,根据斜井施工机械化配套原则,提出高海拔陡坡长斜井机械化安全快速进洞施工工法,并选配超前支护、钻爆、装运、喷锚支护、衬砌等机械化作业线的成套设备。基于高度机械化的设备配套,提出“快挖、快运、快支”的高效施工技术以及各工序快速衔接工艺。以控制围岩变形为核心,构建一套积极干预加固围岩、注重早期支护并快速闭合的主动支护体系。结果表明:与普通钻爆法施工相比,采用三臂凿岩台车高度机械化配套施工效率高,月最快进尺为180 m,较人工钻爆法施工月进尺为130 m相比提高了近30%,施工质量稳定,作业人员投入少,且拱顶沉降和周边收敛均小于规范要求值,衬砌强度可靠,可实现高海拔小断面隧道“快挖、快支、主动支、快封闭”。

联合高阶目标感知与相似匹配的目标跟踪算法

视觉目标跟踪算法利用自注意力机制增强上下文联系,但面对复杂场景时,自注意力机制中的相关性易发生失配,为此提出一种联合高阶目标感知与相似匹配的目标跟踪算法。构建高阶目标感知模型,针对自注意力机制中的一阶自注意图,利用坍塌的极化过滤方式进行空间和通道维度的正交化建模,优化内部相关性;同时组合非线性拟合函数避免坍塌引起的信息损失,进而获得高阶自注意图,捕获具有高阶上下文信息的感知特征。通过不同维度分解目标的感知特征来细化匹配区域,抑制背景噪声并约束当前帧的响应图,提高网络的判别力。在OTB100和UAV123基准的实验结果表明,所提算法有更好的跟踪性能,可以有效应对相似干扰等问题。

网约车合乘均衡匹配与激励策略

针对需求高峰期乘客以红包或调度费等形式激励网约车司机以实现网约车自我调度这一问题,研究了乘客-网约车匹配决策与激励策略选择间的交互关系。基于网约车合乘模式下的均衡匹配状态,以乘客总剩余最大化为目标,考虑匹配、均衡、成本等约束条件,构建了考虑奖励策略的乘客-网约车均衡匹配模型,从乘客视角设计了乘客激励策略、网约车激励策略、乘客和网约车激励策略,并将3种激励策略嵌入到列生成算法对模型进行求解,以实现匹配均衡和价格均衡。以大连市出租车数据为对象进行实证分析,研究结果表明,相比于仅以随机调度费等形式从供给侧激励司机,同时从需求侧和供给侧实施激励策略可以促成更多合乘出行,乘客剩余可提高12.6%;当需求大于供给时,26%的激励在乘客间转移以实现更多的合乘出行;合乘折扣系数也会影响激励流向,同时使用激励策略和合乘折扣策略调节供需可以避免司机恶意竞争和乘客无效激励,在降低乘客出行成本的同时提高司机收益,促成更多合乘出行。

结合注意力机制的卫星遥感影像立体匹配

采用立体匹配技术对多视卫星遥感影像进行三维场景重建一直是摄影测量与遥感领域的核心问题。基于卷积神经网络的深度学习方法极大地促进了立体匹配技术的发展,然而其中涉及匹配困难和误匹配问题的相关研究仍然不足。为了提升卫星遥感影像不适定区域中视差估计的精度,本研究提出了一种结合注意力机制的立体匹配深度学习网络,在特征提取模块中加入注意力机制,分别从通道和空间两个维度捕获全局信息,对特征进行优化;在代价体的构建模块中构建新的代价体积,并重新设置视差的回归范围。为了验证本文方法的有效性,在US3D、WHU-Stereo两个数据集上分别与已有方法Stereo-Net、PSM-Net进行了比较分析。结果表明,本文方法在EPE(endpointerror)和D_1两个指标上均能达到最优,取得了较好的性能,提高了立体匹配的精度,尤其在无纹理、重复纹理、遮挡及视差不连续区域表现出良好的鲁棒性。

子图匹配和强化学习增强的三维点云配准

针对低质量三维点云数据配准精度不足、效率低的问题,为了实现低质量点云的精确、快速配准,提出一种基于子图匹配和强化学习的点云配准方法.首先将三维点云配准转化为一系列离散的刚性变换连续作用结果,利用强化学习策略训练一个端到端的模型以迭代输出刚性变换动作;然后对于模型架构,采用双流主干网络分别提取源点云与目标点云的局部特征信息,设计交叉图注意力模块将源点云图和目标点云图中的相似节点关联起来,使用带选通向量的加权实现图节点的聚合,分别获取源点云图与目标点云图的全局特征表示;最后融合源点云图与目标点云图的全局特征,基于融合特征预测离散的刚性变换动作.强化学习策略的引入显著提高了点云配准算法的泛化性,在加入交叉图注意力模块后,点云配准的精度及效率也进一步被提升.在ModelNet40和ScanObjectNN这2个公共基准数据集上与最新的点云配准方法 ReAgent进行实验的结果表明,所提方法能够将旋转误差的均方差数值降低至少0.16,各向同性旋转误差数值也降低至少0.16,有效地提升低质量点云配准的精度.

基于深度学习与规则结合的海运货物邮件命名实体识别

针对海运货物邮件实体识别中存在识别精度不高、实体边界确定困难的问题,提出一种结合深度学习与规则匹配的识别方法。其中:深度学习方法是在BiLSTM-CRF(Bidirectional Long Short Term Memory-Conditional Random Field)模型的基础上添加词的字符级特征,并融入多头注意力机制以捕获邮件文本中长距离依赖;规则匹配方法则根据领域实体特点制定规则来完成识别。根据货物邮件特点将语料进行标注并划分为:货物名称、货物重量、装卸港口、受载期和佣金五个类别。在自建语料中设置多组对比实验,实验表明所提方法在海运货物邮件实体识别的F1值达到79.3%。