基于等参数线采样的自由曲面优化测量方法

研究了三坐标测量机(CMM)进行自由曲面精度检测时,如何制定采样方案的问题。在已有算法的基础上,提出了两种优化算法:基于法向误差和平均曲率变化的混合采样法,以及自动寻向采样法。前者采样方向给定,由替代曲面与原曲面间的法向误差与曲面平均曲率变化的线性组合决定采样线的位置;后者则依据u、v方向曲率变化大小确定采样线的方向,由最大法向误差确定采样线位置。通过与已有算法进行算例对比,验证了所提方法的有效性。     

基于曲面片拼接的曲面重构算法

通过获取结构表面的正交曲率信息实现板状结构的拟合重构,将板状结构表面分解为曲面片阵列,基于二次方程建立曲面片模型.分析曲面片的边界条件,建立曲面片的非线性方程组.通过逆Broyden秩1法求解非线性方程组,获得曲面片的参数及曲面片方程.通过递推的方法将曲面片进行拼接,实现空间曲面的重构.基于二次方程构建样板曲面,采用数值方法获取非线性方程组的解,并进行仿真实验.仿真实验和数据分析表明,利用已知的正交曲率信息,基于曲面片拼接的曲面重构算法可以有效实现板状结构形变的重构,是一种可行的空间曲面拟合算法.     

点采样曲面的曲率估计

本文提出一种直接在点采样曲面上计算曲面的高斯曲率、平均曲率及主曲率等局部微分性质的方法。首先，为了去除测量产生的噪声和误差，引入高斯核函数为每个采样点加权；接着计算每个采样点的最近邻域点集；然后用双三次B样条曲面最小二乘拟合邻域点集，构造局部参数曲面来逼近原始曲面，以局部参数曲面的曲率估计点采样曲面的曲率；最后给出曲率估计的应用。实验结果表明，该方法能有效地反映曲面的基本特征。     

复杂曲面自动化测量程序生成及分析

着重分析了基于坐标测量机的复杂曲面自动化测量中的测量程序生成和曲面轮廓度评价方法问题。对一张曲面,采用单一测量头,根据曲面的最小曲率半径确定测量头半径,根据曲面精度等级、机床加工能力、曲面的复杂程度等因素确定采样样本大小。利用自适应采样方法,使得采样点的分布完全符合曲率大处采样点多、曲率小处采样点少。评价曲面轮廓度时,针对无基准要求的面轮廓误差采取曲面的最佳匹配,给出了一个参数曲面的最佳匹配算法。     

曲面轮廓度评定技术的研究

本主讨论了曲面轮廓度检测中的两个重要问题——采样点的坐标转换矩阵的求解以及采样点选取方法的研究。提出基于 ｋｒｉｓｔｊａｎ法扩展的牛顿迭代算法来解决曲面检测问题中的采样测量点与设计模型的匹配问题。该方法在计算时间和对初值的敏感性方面,都要优于变分法。同时,以微分几何为基础,根据曲面主曲率与法曲率的关系,采用基于曲面上任一点处主曲率的合成曲率作为曲率测度,来构造反映曲面曲率分布状况的采样形状函数,以它作为采样点选取的评价指标之一。     

曲面边界样点逆向均值漂移识别

针对现有的曲面边界样点识别算法难以适应非均匀分布的实物表面采样数据的问题,将目标样点的k-近邻点集作为曲面局部样本,基于均值漂移算法使得曲面局部样本在一定程度上向目标样点邻近的采样数据稀疏区域扩展,实现对曲面局部样本的增益优化,并对增益优化后的曲面局部样本进行核密度估计,获取目标样点对应的模式点,并通过比较目标样点与其对应模式点的偏离程度进行边界样点判定。实验表明,该算法可快速准确地识别曲面裁剪边界、几何连续的相邻面片公共边界以及曲率变化较大的过渡曲面上的特征样点,并且对非均匀分布的采样数据具有良好的适应性。     

一种数控自动编程直线插补的新算法

本文给出了一种根据零件加工精度的要求 ,对三次B样条列表曲线进行直线插补、自动生成变步长刀位轨迹的算法 ,简称自适应算法。该算法可根据曲线曲率大小 ,自动改变步长 ,同时逼近误差满足要求。该算法稳定、可靠 ,便于在列表曲线、列表曲面的数控加工中推广应用     

离散数据环境下复杂曲面检测点布局策略研究

根据采样策略进行采样点的选择和布局,是对复杂曲面加工质量进行检测轨迹规划的重要内容之一.针对一般曲面采样点布局算法用于离散数据环境下复杂曲面测点布局存在的局限性,通过分析离散数据环境下复杂曲面数据的特点,引入局部网格曲率特性参数,构建基于三角面片微分几何特性的网格简化方法,从而提出基于网格简化的复杂曲面离散数据检测点布局策略,并建立不同层次网格之间距离值相对于简化三角形数量的拟合曲线方程,验证曲率权值优化方法在离散数据环境下对提高复杂曲面采样点布局精度的有效性.     

基于改进遗传算法求取加工自由曲面的最大刀具尺寸

为选取合适的加工刀具尺寸以对自由曲面实现无曲率干涉加工,提出了改进遗传算法并应用于自由曲面全局最大主曲率求取,确定无曲率干涉最大刀具尺寸.根据主曲率特点设计了一种新的非线性自适应度函数,避免算法过早收敛于局部最优值,并给出了新的复制原则.改进遗传算法应用算例分析表明,该算法在500代内能有效求取自由曲面全局最大主曲率,比传统的离散法效率高,选择的最大刀具尺寸满足自由曲面加工无曲率干涉加工的要求.     

改进的重采样算法研究

针对逆向工程中点云数据过多导致拟合曲面光顺性、精度等方面难以达到要求的问题,通过分析对比多种重采样算法的特点,提出一种改进的重采样算法--在区域划分基础上的混合采样法。该算法首先利用曲率估算算法计算各点曲率,然后根据曲率特征将点云数据划分为特征区域和平坦区域,接着对特征区域进行曲率采样,对平坦区域进行均匀采样。通过实验表明,该算法能够有效地减少点云数据,保留细节特征防止出现空白区域,最终获得较高曲面拟合精度。     

A feature-based inspection planning system for coordinate measuring machines

A feature-based inspection planning system is proposed in this research to develop more efficient measuring methodology for a CMM for complicated workpieces having many primitive form features. The proposed strategy is composed of two stages; global inspection planning and local inspection planning stages. In the global inspection planning stage, the system generates an optimum inspection sequence of the features in a part. The sequence is determined by analyzing the feature information such as the nested relations and the possible probe approach directions of the features, and by forming feature groups. A series of heuristic rules are developed to accomplish it. In the local inspection planning stage, each feature is decomposed into its constituent geometric elements, and then the number of sampling points, the locations of the points and the optimum probing sequence are determined. Also, an effective collision avoidance methodology is proposed. After required simulation, the effectiveness of the proposed system is verified .     

Inspection Planning Strategy for the On-Machine Measurement Process Based on CAD/CAM/CAI Integration

The objective of this research is to develop an effective inspection planning strategy for sculptured surfaces in the OMM (on-machine measurement) process. As a first step, effective measuring-point locations are determined to obtain optimum results for given sampling numbers. Two measuring-point selection methods are suggested in this study based on newly proposed CAD/CAM/CAI integration concepts: 1. By the prediction of cutting errors. 2. By considering cutter contact points to avoid the measurement errors caused by cusps. As a next step, the TSP (travelling salesman problem) algorithm is applied to minimise the probe moving distance. Appropriate simulations and experiments are performed to verify the proposed inspection planning strategy in this study, and the results are analysed.     

A computer-aided inspection planning system for on-machine measurement — Part II: Local inspection planning —

As a part II of theis research, new local inspection planning strategy is proposed in this paper based on the proposed inspection feature extraction method. In the local inspection planning stage, each feature is decomposed into its constituent geometric elements for more effective inspection planning. The local inspection planning for the decomposed features are performed to determine: (l) the suitable number of measuring points, (2) their locations, and (3) the optimum probing paths to minimize measuring errors and times. The fuzzy set theory, the Hammersley’s algorithm and the TSP method are applied for the local inspection planning. Also, a new collision checking algorithm is proposed for the probe and/or probe holder based on the Z-map concept. Finally, the results are simulated and analyzed to verify the effectiveness of the proposed methods.     

基于实时重构的自由曲面自适应布点方法

针对现有的自适应算法计算量庞大、在工程中可应用性不强的缺陷,提出在等距法布点的基础上,对测量点进行实时重构来指导三坐标测量机根据曲面本身特性自动增加测量点的方法。测量过程中,当通过拟合五次B样条曲线、三次B样条曲面判断现有测量点不满足精度时,利用两侧测量点的曲率半径,计算出需增加测量点的准确位置,重新拟合包含新检测点的检测样本,直到满足精度要求。以曲面样件为例,辅以计算机图形可视化验证该算法,实验结果表明该方法测量精度及重构精度均可满足数字化检测要求,且在工程应用中相比曲率自适应算法其计算量大幅下降,有效提高了曲面检测效率。     

混合现实装配检测中深度学习的数据增强方法

针对混合现实装配检测中,装配者的视觉检测位姿具有不确定性极易发生误检漏检的问题,提出一种混合现实装配检测中深度学习的数据增强方法。采用人为最佳数据增强策略的数据预处理方法,通过图像增强、几何变换、少量噪声干扰和随机遮挡的方式生成增强数据集,并改善图像增强过程中的特征失真,不仅能有效解决深度学习中人工标注样本任务量大的问题,还有助于提升检测模型的泛化能力。实验结果表明,该方法训练得到的新模型对汽车装配生产线零件的检测精度提升了11.38%。     

Feature selection for automatic visual inspection of solder joints

In this paper, three successive feature reduction methods are employed to select good features for the automatic visual inspection of solder joints. This reduction strategy includes (1) a stability test to remove the features with unstable performance, (2) a separability examination to select the features with good classification capabilities, (3) a correlation analysis to delete the redundant features. Three sets of features are implemented in this feature reduction work: (1) a circular sub-area feature set is related to the intensity conditions within distinct areas in the joint image, (2) a moment of inertia feature set is based upon the intensity of pixels and their relative position in the image plane, (3) a surface curvature feature set analyses the three-dimensional joint topology. Initially 50 features are formulated based on the above strategy. The reduction technique deletes 39 features from this set because of instability, poor performance, and high correlation with other features. Finally, the remaining 11 feature are tested and shown to be superior to state-of-the-art identification methods.     

Integrated precision inspection system for manufacturing of moulds having CAD defined features

An integrated precision inspection system has been developed for manufacturing moulds having CAD defined features. The techniques of precision measurement are implemented for CAD/CAI integration for moulds having sculptured surfaces with some basic features, such as, holes, slots and bosses. Features to be inspected are chosen in the CAD environment, and inspection planning is performed for each feature. The sampling-point strategies are: uniform distribution, curvature dependent distribution, or hybrid distribution of the two depending on the complexity of the sculptured surface. Line and plane features are divided into subintervals, and the measurement points are distributed at random positions in the subinterval. Prime number subintervals are considered for a circle feature, in order to avoid possible periodic distortion of the measurement features. The measurement path planning is performed considering collision avoidance and coordinate matching between the coordinates. The output of the planning is the machine code for a specific CMM having CNC capability. The machine code is downloaded to a specific CMM, and the measurement results are fed back to the computer. A new algorithm, called MINIMAXSURF, is developed to evaluate the form error precisely for sculptured surfaces. The algorithm considers the radius of the touch probe and evaluates the profile tolerance successfully by removing the unavoidable setup errors. The developed measurement system has been applied to real moulds, demonstrating high performance and accuracy.     

基于偏振分像的制冷器件晶粒双面等光程共焦成像缺陷检测装置

针对待测晶粒大的情况,为解决晶粒双面检测时双面之间产生光程差导致的的成像清晰度问题,提出了一种基于偏振分光成像法的半导体制冷器件晶粒相邻双面同时成像缺陷检测的装置。利用偏振分光器与直角转像棱镜对采用偏振分束器与偏振相机的晶粒天面和侧面同时等光程共焦成像检测装置进行了光学设计,完成了晶粒相邻双面同时等光程偏振成像缺陷检测的实验验证。结果表明,该偏振分光成像检测技术可以实现晶粒相邻面的同时缺陷检测,并能很好地满足相邻面等光程成像缺陷检测的性能要求。当晶粒相邻面等光程共焦时,检测分辨率可达到110 lp/mm以上,而当晶粒相邻面离焦（准共焦）仅±0.20 mm时,分辨率则降至45 lp/mm以下。本检测装置具有双面成像清晰度好、成像光路共焦调整方便、检测装置结构简单可靠,以及提高的缺陷检测性能等优点。     

A methodology for smoothing of point cloud data based on anisotropic heat conduction theory

It is necessary to smooth point cloud data in reverse engineering or the inspection of free-form surfaces because noisy points will have a negative influence on the post-processing of this data. The big problem in smoothing point cloud data is how to solve the dilemma between removing noisy points and keeping feature boundary information, whilst controlling the diffusiveness of noisy points. In this paper, the theory of anisotropic heat conduction is adopted to establish a mathematical model of point cloud data smoothing. The point cloud data can be considered as a temperature field with an adiabatic boundary. So the heat is only conducted inside the temperature filed and has no effect on the outer side. For point cloud data, it means that the smoothing is only on the local area, which makes a good balance between deleting noisy points and keeping feature boundary information. The method has been implemented by using two cases for practical application, and the result proves its efficiency.