代数重建和同步迭代重建在电容层析成像中的比较研究

针对电容层析成像技术(ECT)中的逆问题——图像重建算法的非线性和病态性问题,以12电极电容层析成像系统为对象,研究了代数重建算法(ART)与同步迭代重建算法(SIRT).分别对仿真电容值和实测电容值进行图像重建的实验验证,从成像效果、相对误差及重建时间3个方面对这两种算法进行了评估与分析.结果表明,ART算法和SIRT算法均能有效地实现ECT图像重建,其中SIRT算法能够在100次迭代内达到高精度,在收敛速度和成像效果上更具有优势.     

典型场景下EKF-SLAM估计一致性分析

分析了典型场景下基于扩展卡尔曼滤波的同步定位和地图创建(EKF-SLAM)算法的估计一致性.通过理论分析证明了在移动机器人保持静止并持续对一个路标进行观测的场景下,如果机器人的初始位姿协方差矩阵为对角阵,则机器人位置估计的均值和协方差保持不变,而朝向估计将逐步失去一致性.此外,通过蒙特卡罗仿真给出了机器人朝向和路标估计下界的分布情况.结果表明,两者均服从正态分布,因此EKF-SLAM算法在概率意义下给出SLAM系统状态向量的无偏估计.     

单组分体系质量交换网络和换热网络同步综合

为实现单组份体系质量交换网络(mass exchanger network,MEN)和换热网络(heat exchanger network,HEN)同步综合,提出一种新的综合方法,并针对过程系统中质量交换网络和换热网络耦合关系强、连接参数(传质温度)作为优化变量在两网络同步优化时难度大的问题提出了解决方案。采用质量夹点法综合质量交换网络、虚拟温焓图法综合换热网络来构造总网络数学模型,并通过提出连接参数与总网络同时优化的策略,实现质量和换热总网络(com-bined mass and heat exchanger network,CM&HEN)全局同步优化。采用实例对所提方法进行验证,结果表明:该方法所需的年度总费用更低,说明该方法不仅能够实现同步综合,而且能获得更优解。     

Joint segmentation and motion estimation of left ventricle with an anisotropic biomechanical model

Accurate estimation of shape and motion variability of the heart has important clinical implications for heart disease diagnosis. In this paper, we describe a framework which performs simultaneous segmentation and motion analysis of the left ventricle from 3D cardiac images. The left ventricle is formulated as an anisotropic linear elastic body represented by a set of tetrahedrons constructed from sampling nodes. The elastic body is evolved by both internal force and external force, and the external force imposed on each node is an integration of edginess measure, shape coherence measure, prior distributions of the myocardial tissues and temporal driven measure. The displacement field and the segmentation result are simultaneously solved at the equilibrium state between the internal energy and the external energy. The experiment with 3D canine MR images illustrated its abilities and benefits.     

深井孤岛工作面过探巷技术研究

邢东矿2225工作面为孤岛工作面,承受地压大,回采过程中经过2225探巷,巷道围岩破坏剧烈,尤其是底鼓变形严重,初期采取排木垛、打单体等措施在一定程度上控制了探巷顶板下沉,而工作面临近时在煤帮注化学浆形成锚注加固结构使煤体自身强度及承载能力显著增强,锚注加固结构具有较好的整体性和稳定性,从而实现工作面顺利回采。     

解线性方程组的共轭梯度法

通过差分离散Laplace方程,得到一个大型线性方程组,并给出了数值结果。数值实验结果表明共轭梯度法能很好地求解此方程组。     

基于FIG和GWO-SVM的灌浆功率时序预测

为更好地预测灌浆功率时序,建立基于模糊信息粒化(FIG)和灰狼优化支持向量机(GWO-SVM)的灌浆功率时序预测模型。首先,引入信息粒计算方法,将原始详尽的时间序列数值点分解为一系列信息粒,以减少模型的数据输入总量;其次,基于模糊集理论,采用模糊集算子对每个信息粒进行模糊计算,使得到的模糊信息粒可以合理地表示原始数值点集;最后,以支持向量机作为预测工具,并采用灰狼优化算法进行参数寻优,对产生的模糊信息粒进行快速准确的预测。结合实际工程,应用该预测模型对灌浆功率的波动范围和变化趋势进行预测研究,经过性能评价和对比分析,验证了模型的有效性和优越性。     

出行链与出行方式相互影响模式

通过居民出行调查数据分析,建立递归联立离散选择模型,从出行链至出行方式和出行方式至出行链2个方向,重点分析了工作出行链和出行方式间的相互作用.研究表明,模型通过直接量化出行链和出行方式之间的相互作用,加强了行为机制对预测分析的指导.分析结果表明,出行链至出行方式的影响模式在工作出行决策过程中占主导地位,出行者首先根据个人和家庭的需要将一日的活动和出行组织成以出行链为单位的片段,然后在出行链各种性质的约束下考虑选择何种交通方式.研究结果对出行需求预测和管理有指导意义.     

好氧颗粒污泥实现同步除磷脱氮的试验研究

在厌氧反应1h,好氧反应4h,缺氧反应2h的运行条件下,研究了序批式反应器中好氧颗粒污泥同步除磷脱氮的情况,并对好氧颗粒污泥除磷脱氮的机理进行了探讨．试验结果表明,该系统对氮、磷和有机物具有良好的去除效果,对氨氮、总无机氮、磷、COD的去除率分别达到89．2％～98．9％,81．3％～89．4％,86．8％～90．0％和82．7％～96．6％。     

Proposed Methodology for Generation of Building Information Model with Laserscanning

For refurbishment and state review of an existing old building, a new model reflecting the current state is often required especially when the original plans are no longer accessible. Laser scanners are used more and more as surveying instruments for various applications because of their high-precision scanning abilities. For buildings, the most notable and widely accepted product data model is the IFC product data model. It is designed to cover the whole lifecycle and supported by various software vendors and enables applications to efficiently share and exchange project information. The models obtained with the laser scanner, normally sets of points ("point cloud"), have to be transferred to an IFC compatible building information model to serve the needs of different planning states. This paper presents an approach designed by the German Research Center in Karlsruhe (Forschungszentrum Karlsruhe) to create an IFC compatible building information model from laser range images. The methodology through the entire process from data acquisition to the IFC compatible product model was proposed in this paper. In addition, IFC-Models with different level of detail (LoDs) were introduced and discussed within the work.