融合PLS监督特征提取和虚假最近邻点的数据分类特征选择

在高维数据分类中,针对多重共线性、冗余特征及噪声易导致分类器识别精度低和时空开销大的问题,提出融合偏最小二乘(Partial Least Squares,PLS)有监督特征提取和虚假最近邻点(False Nearest Neighbors,FNN)的特征选择方法:首先利用偏最小二乘对高维数据提取主元,消除特征之间的多重共线性,得到携带监督信息的独立主元空间;然后通过计算各特征选择前后在此空间的相关性,建立基于虚假最近邻点的特征相似性测度,得到原始特征对类别变量解释能力强弱排序;最后,依次剔除解释能力弱的特征,构造出各种分类模型,并以支持向量机(Support Vector Machine,SVM)分类识别率为模型评估准则,搜索出识别率最高但含特征数最少的分类模型,此模型所含的特征即为最佳特征子集。3个数据集模型仿真结果:均表明,由此法选择出的最佳特征子集与各数据集的本质分类特征吻合,说明该方法:有良好的特征选择能力,为数据分类特征选择提供了一条新途径。     

A novel offline programming approach of robot welding for multi-pipe intersection structures based on NSGA-Ⅱ and measured 3D point-clouds

The multi-pipe intersection structure in form of co-main pipe is widely used in various industries. To improve its welding quality and efficiency, this paper is devoted to proposing an offline programming approach to the robot trajectory based on NSGA-Ⅱ and measured 3D point-clouds. First, considering the existence of deviation between the actual workpiece and its ideal model, this paper selects the actual 3D point-cloud of weld seam as the research object and extracts its feature points by combining the characteristic of the spatial curve, which can reduce the data density while preserving their geometric characteristics. Second, to ensure the continuity of motion parameters while taking the calculation and fitting accuracy into account, the cubic NURBS is applied to fit the actual weld position, and a fast-adaptive fitting nodes configuration scheme is designed according to the variation characteristics of the spatial curve and the fitting error restriction. Third, this paper introduces a trajectory adaptive discretization method based on the chord error constraint for robot program generation, and establishes an optimization model of the robot welding trajectory based on NSGA-Ⅱ frame, which gives the robot joint's motion trajectories with optimal welding quality and cable twisting fluctuation. Finally, the experiments are designed to verify the correctness of the aforementioned approach.     

基于特征指数加权的最小二乘支持向量机算法

根据支持向量回归机原理,针对样本特征对回归预测重要性的差异,采用最小二乘支持向量回归机(LS-SVR)算法,减少参数数量,针对参数对预测效果的影响,并考虑到特征加权的意义,采用特征指数进行加权,其权重系数由灰色关联度确定,提出了基于特征指数加权的最小二乘支持向量回归机算法。为验证该算法的有效性,对实际股票价格进行预测,结果表明该算法较传统最小二乘支持向量回归机算法,其回归估计函数的预测能力明显提高,具有一定的实用价值。     

基于特征相关的偏最小二乘特征选择方法

针对传统的偏最小二乘法只考虑单特征的重要性以及特征之间存在冗余和多重共线性等问题，将特征之间的统计相关性引入到传统的偏最小二乘分析中，构造了一种基于特征相关的偏最小二乘模型。首先利用特征相关度对特征进行评估预选出特征组，然后将其放入偏最小二乘模型中进行训练，评估该特征组是否可取。结合前向贪心搜索策略依次评价候选特征，并选中使目标函数最小的候选特征加入到已选特征。分别采用麻杏石甘汤君药止咳、平喘和UCI数据集进行分析处理，实验结果表明，该特征选择方法能较好寻找较优的特征组。     

基于最小平方中值定理的立体视觉里程计

提出了一种基于最小平方中值定理（LMedS）的立体视觉里程计方法。利用图像中尺度不变的SIFT特征点作为路标,基于KD树的最邻近点搜索算法来实现左右图像对特征点的匹配和前后帧间特征点跟踪。通过特征点的三维重建,基于最小平方中值定理估计出机器人的运动距离和方向信息。实验表明该方法在不同图像间匹配、三维路标跟踪和机器人运动估计中具有很强的鲁棒性。     

A surface based approach to recognition of geometric features for quality freeform surface machining

The paper presents a surface-based approach for geometric feature recognition for the purpose of automating the process planning of freeform surface machining. The proposed approach consists of the following four steps for recognition of the geometric features: conversion and preprocessing of the surface geometry data, subdivision of NURBS surface, reconstruction of surface orientation areas, and recognition of geometric features. The proposed scheme assumes that the input geometry data form is based on an IGES CAD model and the surface model can be represented in the form of trimmed NURBS surfaces. The connectivity relations of the product surfaces are analyzed and each surface is subdivided into orientation regions based on the surface normal vector over a certain point density grid, and then all the connected regions with the same orientation type are grouped into surface orientation areas. After that, the geometric feature will be recognized through the analysis of area connectivity and relationship. The paper describes the developed algorithms on surface orientation region subdivision, surface orientation area reconstruction, and geometric feature recognition. The verified feasibility study of the developed method is also presented.     

Dealing with feature interactions for prismatic parts in STEP-NC

Determining the precedence of machining features is a critical issue in feature-based process planning. It becomes more complex when geometric interaction occurs between machining features. STEP-NC, the extension of STEP (ISO 10303) standard developed for CNC controllers, is a feature-based data model. It represents all the geometric and topological product data minus feature interactions. In this paper, machining precedence of interactive and non-interactive STEP-NC features is discussed. Local and global precedence of machining features are defined on the basis of geometric constraints, such as geometric interaction of features and feature approach face and technological constraint such as access direction of the cutting tool. A software tool has been developed to visualize the STEP-NC part model and to generate the graphs of feature interaction and feature precedence. The output can be then used to augment the STEP-NC data in order to generate the optimal sequence of operations.     

一种基于联合轮廓特征矢量的目标识别方法

针对基于形状特征进行目标识别的方法存在的不足,提出一种联合轮廓不变矩特征和轮廓几何特征的识别方法;针对不同的待识别目标,采集足够数量的训练样本,统计每一个轮廓特征的均值、标准差和变异系数;并据此对这些轮廓特征进行动态筛选和加权,建立起待识别目标的联合轮廓特征矢量模型.在线识别时,提取场景目标的联合轮廓特征矢量,对其进行...     

基于熵值法的加权最小二乘支持向量机

支持向量机是一种以统计学习理论为基础的机器学习算法,着重解决小样本的建模问题,并且对非线性高维数据具有较好的处理能力。通常对于多维特征的数据,会对每一维数据做归一化处理以消除量纲的影响,但缺点在于忽视了各维特征的权重差异。提出了一种加权最小二乘支持向量机的建模方法,通过熵值法确定每一维特征的权重,根据特征权重对数据进行加权处理,最后由最小二乘支持向量机建立该系统模型。实验表明,对于多维特征的数据,所提方法具有更好的建模效果。     

基于偏最小二乘回归的人脸身份和表情同步识别方法

将偏最小二乘回归方法用于人脸身份和表情的同步识别。首先,对每幅人脸图像进行脸部特征提取以及相应的语义特征定义。在脸部特征提取方面,从每幅图像中标定出若干脸部关键点位置,并提取图像在该关键点处的Gabor小波系数（Gabor特征）以及关键点的坐标值（几何特征）,作为该图像的输入特征。语义特征则定义为该人脸图像所属的表情类别信息以及所对应的人脸身份信息。其次,利用核主成分分析(KPCA)方法对脸部Gabor特征和几何特征进行融合,使得输入特征具有更好的识别特性;最后,运用偏最小二乘回归(PLSR)方法建立脸部特征和语义特征之间的关系模型,并运用此模型对某一测试人脸图像进行表情和身份的同步识别。通过在JAFFE国际表情数据库和AR人脸数据库上的对比实验,证实了所提方法的有效性。     

一种模拟电路的支持向量机故障诊断方法

系统地提出了模拟电路的最小二乘小波支持向量机故障诊断方法。从测试点得到各种故障状态下的输出电压信号，对输出电压信号进行小波去噪，对信号进行小波分解获取多尺度的低频系数和高频系数，并对小波系数进行处理从而提取出故障特征量，以此作为学习样本来训练最小二乘小波支持向量机，确定其模拟电路故障诊断的模型。雷达系统电路仿真结果表明了模拟电路的小波变换和最小二乘小波支持向量机故障诊断方法取得了较好的效果。     

Entropy and Cross‐entropy for Generalized Hesitant Fuzzy Information and Their Use in Multiple Attribute Decision Making

In this paper, we present the entropy, cross‐entropy, and similarity measure for generalized hesitant fuzzy information and discuss their desirable properties. Some measure formulas are developed, and the relationships among them are investigated. We show that the similarity measure and entropy for generalized hesitant fuzzy information can be transformed by each other based on their axiomatic definitions. Then we develop two approaches for solving multiple attribute decision making, in which the attribute values are given in the form of generalized hesitant fuzzy elements (GHFEs). In the first approach, the attribute weight vector is determined by the generalized hesitant fuzzy entropies, and the optimal alternative is obtained by comparing the generalized hesitant fuzzy cross‐entropies between alternatives and positive‐ideal or negative‐ideal solutions; in the second approach, the attribute weight vector is derived from the maximizing deviation method and optimal alternative is obtained by using the technique for order preference by similarly to ideal solution (TOPSIS) method. Finally, an example is provided to illustrate the practicality and effectiveness of the developed approaches.     

A novel geometric feature extraction method for ear recognition

The discriminative ability of geometric features can be well supported by empirical studies in ear recognition. Recently, a number of methods have been suggested for geometric feature extraction from ear images. However, these methods usually have relatively high feature dimension or are sensitive to rotation and scale variations. In this paper, we propose a novel geometric feature extraction method to address these issues. First, our studies show that the minimum Ear Height Line (EHL) is also helpful to characterize the contour of outer helix, and the combination of maximal EHL and minimum EHL can achieve better recognition performance. Second, we further extract three ratio-based features which are robust to scale variation. Our method has the feature dimension of six, and thus is efficient in matching for real-time ear recognition. Experimental results on two popular databases, i.e. USTB subset1 and IIT Delhi, show that the proposed approach can achieve promising recognition rates of 98.33% and 99.60%, respectively.     

A new approach to geometric nonlinearity of cable structures

The basic structural principles surrounding nonlinear behaviour of cable networks are explained through the example of a two-link structure. The nonlinear static response to load for this structure is then derived explicitly using the proposed simple approach, and results are compared with those obtained from a general two-dimensional non-linear bar element (derivation given), and to results quoted in the literature. The proposed approach to geometric nonlinearity is then tested on three three-dimensional cable networks and the results compared with those obtained by three other techniques, namely geometric stiffness matrix, dynamic relaxation and general minimum energy. The proposed technique has been found to be comparable to established techniques in accuracy, stability and speed of solution while at the same time exhibiting the key features of separation of the numerical computation from the underlying structural mechanics, and the requirement of understanding only the most elementary of structural mechanics. The proposed technique is thus also most suitable for introducing cable structures to undergraduate courses.     

基于Laplacian正则化最小二乘的半监督SAR目标识别

提出了一种基于核主成分分析(kernel principal component analysis,简称KPCA)和拉普拉斯正则化最小二乘(Laplacian regularized least squares,简称LapRLS)的合成孔径雷达(synthetic aperture radar,简称SAR)目标识别方法.KPCA特征提取方法不仅能够提取目标主要特征,而且有效地降低了特征维数.Laplacian正则化最小二乘分类是一种半监督学习方法,将训练集样本作为有标识样本,测试集样本作为无标识样本,在学习过程中将测试集样本包含进来以获得更高的识别率.在MSTAR实测SAR地面目标数据上进行实验,结果表明,该方法具有较高的识别率,并对目标角度间隔具有鲁棒性.与模板匹配法、支撑矢量机以及正则化最小二乘监督学习方法相比,具有更高的SAR目标识别正确率.此外,还通过实验分析了不同情况下有标识样本数目对目标识别性能的影响.     

VM-TEST: Mechanical property measurement using electrostatically actuated vertical MEMS test structures fabricated in thick metal layers

An efficient method is presented to determine the mechanical properties of thick metal layers using the pull-in voltage of electrostatically actuated structures. To fabricate these high aspect ratio beams without severe deformations, additional features were added, which made existing pull-in voltage equations inaccurate and therefore corrections were necessary. ANSYS Multiphysics was used to analyze the differences between ideal beams and the fabricated beams. To demonstrate the proposed approach, both nickel and gold devices were fabricated. To extract the material property values, a sum of least squares fitting scheme was used. A Young’s modulus of 186.2 and 60.8 GPa was obtained for nickel and gold structures respectively. Both values are significantly smaller than values reported for bulk material, but fall within the range of values reported in the literature.     

基于多节点信息融合和协作的信号调制方式识别

针对单节点在低信噪比环境下调制识别率低的难题,提出了基于一种多节点信息融合和协作的信号调制方式识别方法。首先设计多个传感器节点协作方案,并提取每节点特征,然后中心节点将各节点特征进行融合,最后采用最小二乘支持向量机建立信号调制分类器。仿真结果表明,相比于其他信号调制识别方法,该方法提高了信号调制识别精度,对信噪比环境具有更好的自适应性。     

Algorithmic optimization method for effort estimation

This paper presents an improvement of an effort estimation method that can be used to predict the level of effort for software development projects. A new estimation approach based on a two-phase algorithm is used. In the first phase, we apply a calculation based on use case points (UCPs). In the second phase, we add correction values (a ₁, a ₂) obtained via least squares regression. This approach employs historical project data to refine the estimate. By applying the least squares regression approach, the algorithm filters out estimation errors caused by human factors and company practice.     

霍尔式微位移传感器建模的实现方法

为了正确反映霍尔式微位移传感器的特性,本文首先介绍霍尔式微位移传感器的工作原理,得出霍尔式微位移传感器被测试件位移量与相关测量电路输出电压(S,V)关系特征,然后基于最小二乘估计算法基本原理,提出了运用MATLAB语言建立霍尔式微位移传感器(S,V)关系特征的数学模型的方法,给出建模的程序流程图以及仿真结果。