面向协同设计的VRML模型装配特征信息重建方法

VRML模型具有适用于网络化协同设计的许多优点,但在应用时,其离散的三角面片构型造成了大量设计信息的缺失,限制了协同设计操作的实现。由于机械设计中回转体约占70%,其特征参数（如回转中心,回转半径等）在设计过程中有重要意义。为此提出了基于向量空间的回转体特征参数提取方法,对装配特征信息进行重建,为基于VRML模型的协同浏览、尺寸信息查询、模型装配定位等操作提供重要参数。对算法复杂度进行了分析。实验表明在高精度要求下算法具有很高的效率。     

提高小样本高光谱图像分类性能的变维卷积神经网络

目的 为了解决基于卷积神经网络的算法对高光谱图像小样本分类精度较低、模型结构复杂和计算量大的问题,提出了一种变维卷积神经网络。方法 变维卷积神经网络对高光谱分类过程可根据内部特征图维度的变化分为空—谱信息融合、降维、混合特征提取与空—谱联合分类的过程。这种变维结构通过改变特征映射的维度,简化了网络结构并减少了计算量,并通过对空—谱信息的充分提取提高了卷积神经网络对小样本高光谱图像分类的精度。结果 实验分为变维卷积神经网络的性能分析实验与分类性能对比实验,所用的数据集为Indian Pines和Pavia University Scene数据集。通过实验可知,变维卷积神经网络对高光谱小样本可取得较高的分类精度,在Indian Pines和Pavia University Scene数据集上的总体分类精度分别为87.87%和98.18%,与其他分类算法对比有较明显的性能优势。结论 实验结果表明,合理的参数优化可有效提高变维卷积神经网络的分类精度,这种变维模型可较大程度提高对高光谱图像中小样本数据的分类性能,并可进一步推广到其他与高光谱图像相关的深度学习分类模型中。     

Structural modelling and integrative analysis of microelectromechanical systems product using graph theoretic approach

Micro-electromechanical systems (MEMS) as an enabling technology is seen to play a more and more important role for the main stream of industry of the future by broadening its applications to information, communications and bio technologies. Development of MEMS devices, however, still relies on knowledge and experience of MEMS experts due to the design and fabrication process complexity. It is difficult to understand the trade-offs inherent in the system and achieve an optimal structure without any MEMS-related insight. An attempt is made to develop an integrated systems model for the complete structure of the MEMS product system in terms of its constituents and interactions between the constituents. The hierarchical tree structures of the MEMS system and its subsystems are presented up to component level. For characterization, analysis and identification of MEMS product system, three different mathematical models say graph theoretic model, matrix model and permanent model are presented. These models are associated with graph theory, matrix method and variable permanent function by considering the various subsystems, subsubsystems up to component level, their connectivity and interdependency of the MEMS product system. The developed methodology is explained with an example. The proposed modeling and analysis is extendable to the subsystems and the component level. An overall structural analysis can be carried out by following a ‘top-down’ approach or ‘bottom-up’ approach. Understanding of MEMS product structure will help in the improvement of performance, cost, design time, and so on.     

A computer-aided design system for framed-mould in autoclave processing

The general computer-aided design (CAD) software cannot meet the mould design requirement of the autoclave process for composites, because many parameters such as temperature and pressure should be considered in the mould design process, in addition to the material and geometry of the part. A framed-mould computer-aided design system (FMCAD) used in the autoclave moulding process is proposed in this paper. A function model of the software is presented, in which influence factors such as part structure, mould structure, and process parameters are considered; a design model of the software is established using object oriented (0-0) technology to integrate the stiffness calculation, temperature field calculation, and deformation field calculation of mould in the design, and in the design model, a hybrid model of mould based on calculation feature and form feature is presented to support those calculations. A prototype system is developed, in which a mould design process wizard is built to integrate the input information, calculation, analysis, data storage, display, and design results of mould design. Finally, three design examples are used to verify the prototype.     

异常步态3维人体建模和可变视角识别

目的 运用视觉和机器学习方法对步态进行研究已成为当前热点,但多集中在身份识别领域。本文从不同的视角对其进行研究,探讨一种基于点云数据和人体语义特征模型的异常步态3维人体建模和可变视角识别方法。方法 运用非刚性变形和蒙皮方法,构建基于形体和姿态语义特征的参数化3维人体模型;以红外结构光传感器获取的人体异常步态点云数据为观测目标,构建其对应形体和姿态特征的3维人体模型。通过ConvGRU（convolution gated necurrent unit）卷积循环神经网络来提取其投影深度图像的时空特征,并将样本划分为正样本、负样本和自身样本三元组,对异常步态分类器进行训练,以提高分类器对细小差异的鉴别能力。同时对异常步态数据获取难度大和训练视角少的问题,提出了一种基于形体、姿态和视角变换的训练样本扩充方法,以提高模型在面对视角变化时的泛化能力。结果 使用CSU（Central South University）3维异常步态数据库和DHA（depth-included human action video）深度人体行为数据库进行实验,并对比了不同异常步态或行为识别方法的效果。结果表明,本文方法在CSU异常步态库实验中,0°、45°和90°视角下对异常步态的综合检测识别率达到了96.6%,特别是在90°到0°交叉和变换视角实验中,比使用DMHI（difference motion history image）和DMM-CNN（depth motion map-convolutional neural network）等步态动作特征要高出25%以上。在DHA深度人体运动数据库实验中,本文方法识别率接近98%,比DMM等相关算法高出2%~3%。结论 提出的3维异常步态识别方法综合了3维人体先验知识、循环卷积网络的时空特性和虚拟视角样本合成方法的优点,不仅能提高异常步态在面对视角变换时的识别准确性,同时也为3维异常步态检测和识别提供一种新思路。     

Mapping 2D midship drawings into a 3D ship hull model based on STEP AP218

Standard for the exchange of product model data (STEP) AP218 has a standard schema to represent the structural model of a midship section. While its schema facilitate the exchange of ship structural models among heterogeneous systems, most shipyards and classification societies exchange information using 2D paper drawings. We propose a feature parameter input method to generate a 3D STEP model of a ship structure from 2D drawings. We have analyzed the ship structure information contained in 2D drawings and have defined a data model to express the contents of the drawing. We also developed a GUI for the feature parameter input. To translate 2D information extracted from the drawing into a STEP AP218 model, we have developed a shape generation library and have generated the 3D ship model through this library. The generated 3D STEP model of a ship structure can be used to exchange information between design departments in a shipyard as well as between classification societies and shipyards.     

ArcGIS实现三维燃气管网仿真的研究

三维GIS能够较准确和高效地实现燃气管网的设计和管理,由此探讨了ArcGIS实现三维燃气管网的仿真.从数字高程和数字地形三维建模生成周边环境,到应用三维管线设计技术生成管网矢量数据,介绍了三维管线仿真环境建模原理,以及使用ArcGIS实现三维燃气管网仿真的系统层次结构和功能扩展,并且应用实例,展示系统设计开发的可行性和实际应用效果.     

Micro harvester using isotropic charging of electrets deposited on vertical sidewalls for conversion of 3D vibrational energy

The design and fabrication process of an integrated micro energy harvester capable of harvesting electrical energy from low amplitude mechanical vibrations is presented. A specific feature of the presented energy harvester design is its capability to harvest vibrational energy from different directions (3D). This is done through an innovative approach of electrets placed on vertical sidewalls, allowing miniaturization of 3D capacitive energy harvester fabrication on monolithic CMOS substrates. A new simple electret charging method using ionic hair-dryers is used. The charging performance of SiO₂ and CYTOP electrets are characterized for electrets in horizontal arrangement and electrets deposited on vertical sidewalls.     

高光谱图像小样本分类的卷积神经网络方法

目的 与传统分类方法相比,基于深度学习的高光谱图像分类方法能够提取出高光谱图像更深层次的特征。针对现有深度学习的分类方法网络结构简单、特征提取不够充分的问题,提出一种堆叠像元空间变换信息的数据扩充方法,用于解决训练样本不足的问题,并提出一种基于不同尺度的双通道3维卷积神经网络的高光谱图像分类模型,来提取高光谱图像的本质空谱特征。方法 通过对高光谱图像的每一像元及其邻域像元进行旋转、行列变换等操作,丰富中心像元的潜在空间信息,达到数据集扩充的作用。将扩充之后的像素块输入到不同尺度的双通道3维卷积神经网络学习训练集的深层特征,实现更高精度的分类。结果 5次重复实验后取平均的结果表明,在随机选取了10%训练样本并通过8倍数据扩充的情况下,Indian Pines数据集实现了98.34%的总体分类精度,Pavia University数据集总体分类精度达到99.63%,同时对比了不同算法的运行时间,在保证分类精度的前提下,本文算法的运行时间短于对比算法,保证了分类模型的稳定性、高效性。结论 本文提出的基于双通道卷积神经网络的高光谱图像分类模型,既解决了训练样本不足的问题,又综合了高光谱图像的光谱特征和空间特征,提高了高光谱图像的分类精度。     

Range identification for nonlinear parameterizable paracatadioptric systems

In this paper, a new range identification technique for a calibrated paracatadioptric system mounted on a moving platform is developed to recover the range information and the three-dimensional (3D) Euclidean coordinates of a static object feature. The position of the moving platform is assumed to be measurable. To identify the unknown range, first, a function of the projected pixel coordinates is related to the unknown 3D Euclidean coordinates of an object feature. This function is nonlinearly parameterized (i.e., the unknown parameters appear nonlinearly in the parameterized model). An adaptive estimator based on a min-max algorithm is then designed to estimate the unknown 3D Euclidean coordinates of an object feature relative to a fixed reference frame which facilitates the identification of range. A Lyapunov-type stability analysis is used to show that the developed estimator provides an estimation of the unknown parameters within a desired precision. Numerical simulation results are presented to illustrate the effectiveness of the proposed range estimation technique.     

字典神经网络方法快速湍流烟雾合成

目的 基于物理的烟雾模拟是计算机图形学的重要组成部分,渲染具有细小结构的高分辨率烟雾,需要大量的计算资源和高精度的数值求解方法。针对目前高精度湍流烟雾模拟速度慢,仿真困难的现状,提出了基于字典神经网络的方法,能够快速合成湍流烟雾,使得合成的结果增加细节的同时,保持高分辨率烟雾结果的重要结构信息。方法 使用高精度的数值仿真求解方法获得高分辨率和低分辨率的湍流烟雾数据,通过采集速度场局部块及相应的空间位置信息和时间特征生成数据集, 设计字典神经网络的网络架构,训练烟雾高频成分字典预测器,在GPU（graphic processing unit）上实现并行化,快速合成高分辨率的湍流烟雾结果。结果 实验表明,基于字典神经网络的方法能够在非常低分辨率的烟雾数据下合成空间和时间上连续的高分辨率湍流烟雾结果,效率比通过在GPU平台上直接仿真得到高分辨率湍流烟雾的结果快了一个数量级,且合成的烟雾结果与数值仿真方法得到的高分辨率湍流烟雾结果足够接近。结论 本文方法解决了烟雾的上采样问题,能够从非常低分辨率的烟雾仿真结果,通过设计基于字典神经网络结构以及特征描述符编码烟雾速度场的局部和全局信息,快速合成高分辨率湍流烟雾结果,且保持高精度烟雾的细节,与数值仿真方法的对比表明了本文方法的有效性。     

Linewidth design rules for microstructures manufactured in PMMA resist using X-ray LIGA

During the past decade, individual simulation modules for the LIGA process have been developed, such as the power spectrum of the X-rays available from a given synchrotron, the effect of various components in the beam line, the image structure in the resist and image development. This has lead to an understanding of the parameters affecting the basic dimensional relationship between the mask and the 3D image in the resist itself. The commercialization of X-ray LIGA now requires knowledge of the parameters affecting the production engineering of micro devices. Foremost among these parameters are the manufacturing tolerances arising from variations in key microfabrication steps. This paper seeks to integrate the various physical simulation models into a single interactive program, whereby the design engineer can see the effect of different synchrotron output spectra, different beamline constructions and different resist parameters on the dimensions of the planned 3D microstructure. A novel approach to estimating the dimensional tolerances to be expected with large scale production is developed and comparison made with data available from the literature. Predicted results from using a particular synchrotron with a beamline designed to approximate to a “standard” exposure spectra at the rest surface are calculated, showing the microstructure shape and tolerance expected in large scale production. From these data a basic set of equations are established from which “Linewidth Design rules for LIGA” can be established. The steps following definition of the resist structure, for example electrodeposition, injection moulding or hot embossing, are not studied here.     

Multi-agent collaborative 3D design with geometric model at different levels of detail

This paper presents a scheme for collaborative 3D design using product model at various levels of detail (LODs). Design features are selectively hidden at each level from certain participants, depending on their actual needs and individual accessibility in the collaboration. A tree data structure represents the feature hierarchy of CAD construction, the link between feature and LOD, and 2D mesh data for display control of each feature. An XML/XSLT-based approach is proposed to enable real-time visualization of different LOD models in distributed environment. A collaborative design system is implemented using multi-agent technologies, which focuses on function design of each agent, interactions among agents, the client–server structure, and generation of the LOD data using the XML/XSLT approach. A scenario of synchronous 3D mold assembly demonstrates that geometric categorization of product model provides an operational mechanism for assuring security of information sharing in engineering collaborations over the Internet. It also validates the effectiveness of the agent technologies for automating complex engineering activities.     

Design and optimization of a tunable W‐band subharmonic cavity based Gunn diode oscillator using computer‐aided design technique

In this work, a systematic computer‐aided design technique is proposed to minimize the fabrication iteration for the design and development of W‐band subharmonic Gunn diode oscillator with wideband tunable bandwidth at W‐band. Gunn diode based single diode oscillator structure was divided into passive and active parts to facilitate the modeling of the component on appropriate simulation environment. Resonating structure and package of Gunn diode are modeled as passive circuit in high frequency structure simulator (HFSS). To satisfy the oscillator design equation, disc‐post resonating structure is tuned in HFSS and its S‐parameters are collaborated with the model of Gunn diode in advanced design system. Magnitude and phase of reflection coefficient (S11) is observed to ascertain the desired frequency of oscillation. Proper tuning of disc‐post structure is done on simulation platform, which reduces the fabrication complexity and cost as well. The measurement results validate the models designed using EM and circuit simulator. The measured maximum stable RF power without any fabrication iteration is 14.2 dBm. A tunable bandwidth of 4 GHz with power output ripple of ±1 dB is measured by using a movable backshort.     

Feature-based 3D non-manifold freeform object construction

This paper presents a novel technique for modeling a 3D non-manifold freeform model around a 3D reference model. To represent both the design abstractions and the incomplete topological information, a new non-manifold data structure is first defined. Our data structure embodies the functional vitalities of both the boundary representation data structure and the complex-based data structure. Along with our data structure, a set of topological operators is defined to manipulate the entities in the data structure. Based on the non-manifold data structure and the topological operators, we develop a technique to construct 3D freeform objects around a reference model. Intuitive 2D sketches are adopted to specify the detailed profile of the object constructed. The construction method is feature-based – every reference model has pre-defined features, and the feature template of the constructed object is related to the features of the reference model by feature node encoding. Therefore, the surfaces derived from one reference model can be regenerated automatically on another reference model with the same features. The geometry coverage of our geometric modeling approach includes both manifold and non-manifold 3D freeform objects.     

基于特征单元的NC过程分析及信息建模

文章通过对2.5轴和3轴铣削加工过程及信息需求分析,详细论述了基于NC特征单元的数控加工过程。并在此基础上构造了以NC特征单元为核心的NC领域产品信息模型。该产品模型不仅适合于CAD/CAPP/NCP集成,而且能够满足NCP/CNC的信息集成需求。     

An initial hull structural modeling system for computer-aided process planning in shipbuilding

At the initial stage of ship design, it is difficult for designers to define all design information of a hull structure on 2D drawings. Thus, other designers must undertake the arduous task of translating such information to generate a 3D CAD model of the hull structure which is required at the following design stage such as the initial process planning stage. Since this task needs much time and effort, the 3D CAD model is not being generated at the initial design stage. For solving this problem, an initial hull structural modeling system is developed in this study. The applicability of the developed system is demonstrated by applying it to a deadweight 300,000 ton VLCC (Very Large Crude oil Carrier). The results show that the developed system can quickly generate the 3D CAD model of the hull structure and accurately extract the production material information for computer-aided process planning at the initial design stage.