基于风格迁移的交互式航空发动机孔探图像扩展方法

在航空发动机孔探图像缺陷检测任务中，缺陷区域样本数量远少于正常样本数量，且缺陷样本无法覆盖整个样本空间，导致检测算法泛化能力较差。针对上述问题，提出了一种基于风格迁移技术的交互式数据扩展方法。首先，通过交互界面选择背景图像和缺陷目标，并根据背景图像指定需要粘贴的目标的大小、角度和位置等信息；其次，通过风格迁移技术将背景图像的风格迁移到目标图像上，使得背景图像和待检测目标具有相同的风格；最后，利用泊松融合算法对融合区域的边界进行修正，以达到连接区域自然过渡的效果。通过二分类和缺陷检测验证了该方法的有效性。在包含真实图像和扩展后图像的二分类实验中，测试人员的平均分类错误率达到了44.0%；而在基于Mask R-CNN模型的检测任务中，所提方法的检测和分割平均精度（AP）较传统方法分别提高了99.5%和91.9%。     

基于风格迁移的交互式航空发动机孔探图像扩展方法

在航空发动机孔探图像缺陷检测任务中,缺陷区域样本数量远少于正常样本数量,且缺陷样本无法覆盖整个样本空间,导致检测算法泛化能力较差。针对上述问题,提出了一种基于风格迁移技术的交互式数据扩展方法。首先,通过交互界面选择背景图像和缺陷目标,并根据背景图像指定需要粘贴的目标的大小、角度和位置等信息;其次,通过风格迁移技术将背景图像的风格迁移到目标图像上,使得背景图像和待检测目标具有相同的风格;最后,利用泊松融合算法对融合区域的边界进行修正,以达到连接区域自然过渡的效果。通过二分类和缺陷检测验证了该方法的有效性。在包含真实图像和扩展后图像的二分类实验中,测试人员的平均分类错误率达到了44.0%;而在基于Mask R-CNN模型的检测任务中,所提方法的检测和分割平均精度（AP）较传统方法分别提高了99.5%和91.9%。     

Feedback strategies for visual search in airframe structural inspection

Feedback of information has consistently shown positive results in human inspection, provided it is given in a timely and appropriate manner. Feedback serves as the basis of most training schemes; traditionally this has been performance feedback. Other forms of feedback which provide strategy information rather than performance information may have a role in improving inspection. This study compared performance feedback and cognitive feedback in a realistic simulation of an aircraft structural inspection task. Performance (time, errors) feedback showed the greatest improvements in performance measures. Cognitive feedback enhanced efficiency measures of search strategy. When cognitive feedback consisted of visual representations of the path and the coverage of the search sequence, subjects also were able to use this task information to improve their search performance.     

基于特征部位圆形域的人脸图像修复方法

针对基于样本块的纹理合成方法存在的修复结构不合理和效率较低的问题,提出基于特征部位圆形域的人脸图像修复方法。首先进行人脸特征点定位,依据特征点分布将人脸图像分割出4个特征部位圆形域,明确特征搜索域范围。然后在优先级模型中以指数函数的形式改变置信度项的衰减趋势,并结合结构梯度项使用局部梯度信息约束优先级,提高修复结果的结构连通性。在匹配块搜索阶段,根据目标块与各个特征部位圆形域的相对位置,确定匹配块的搜索域,提升搜索效率。最终在结构相似性的标准下选择结构最佳匹配块,完成结构连通的人脸图像修复。与4个先进的方法相比较,所提方法修复图像的峰值信噪比（PSNR）平均提升了1.219~2.663 dB,时间消耗平均减小了34.7%~69.6%。实验结果表明,该方法对保持人脸图像的结构连通性和视觉合理性有显著效果,在修复的精度和时间上都表现优异。     

基于随机特征映射的四层多核学习方法

针对单核网络模型的核函数选择无理论依据以及基于随机特征映射的四层神经网络(FRMFNN)节点规模过大的问题,提出了一种基于随机特征映射的四层多核学习神经网络(MK-FRMFNN)算法.首先,把原始输入特征通过特定的随机映射算法转化为随机映射特征;然后,经过不同的随机核映射生成多个基本核矩阵;最后,将基本核矩阵组成合成核...     

Path planning method for on-machine inspection of aerospace structures based on adjacent feature graph

On-Machine Inspection (OMI) system has been playing an important role in modern aeronautical manufacturing, owing to its high efficiency, great convenience, and low application cost. The inspection path planning of OMI, which aims to find an optimal path that traverses all inspection points without collisions and by costing as little time as possible, is a main bottleneck that limits the achievements of higher efficiency and less inspection time. Besides, the path planning for OMI in aerospace manufacturing faces new challenges, which are barely explored, because aerospace structures have unique characteristics of complex geometrical features, large-scale dimensions and high-precision processing requirements. Thus, this paper proposes a novel, easily-implemented and robust inspection path planning method to plan paths for OMI of aerospace structures based on the properties of aerospace structures. In order to lift the inspection efficiency, this method makes three improvements on the path planning. First, reorganize the inspection features based on the cluster technology. Second, construct the adjacent feature graph based on Voronoi Diagram to plan the path. Third, a search algorithm is designed to search the adjacent feature graph to decide the sequence of inspection features and a convex hull based algorithm is used to avoid collisions. The proposed method has been tested for several cases and solid experimental results have shown that these improvements take effects in path planning for OMI of aerospace structures and suited paths can be provided for the inspection.     

基于特征点约束的人脸纹理映射

人脸纹理映射技术是计算机辅助颅骨面貌复原中一种特殊的真实感处理技术。针对人脸面部器官纹理映射难以准确实现的问题,提出一种基于特征点约束的人脸纹理映射方法。利用最小二乘保角映射参数化时需固定顶点来完成特征点约束。通过对大量单张、正面照片作为纹理进行映射,证实了该方法能够取得良好的映射效果。实验结果表明本方法鲁棒且效率高,降低了算法的复杂性。     

一种适用于零件批量检测的尺寸批量测量方法

对于工业零件的检测,目前主要采用人工抽检的接触式尺寸测量或基于机器视觉的非接触式测量。针对零件批量检测时需人工逐一替换检测对象、测量过程耗时长等问题,提出了一种适用于零件批量检测的尺寸批量测量方法,包含零件识别、待测量区域定位、边缘检测与最小二乘拟合、基于多线程的并行测量等主要流程。实验结果表明,测量精度误差不超过2.1μm,重复性测量误差不超过0.8μm,对于包含100个手工检测合格的工业常用圆角小方块零件的5528×10416像素拼接图像,测量速度不超过9s,批量检测结果均为合格(公差为3.0μm),能满足工业批量检测的需求,有效解决工业实际问题。     

Automated inspection process planning: Algorithmic inspection feature recognition, and inspection case representation for CBR

General metrological inspection planning is among the least explored computer-aided process planning (CAPP) domains. This paper explores certain basic issues involved in inspection planning using case-based reasoning in an environment of a Generic CAPP Support System. Firstly, algorithmic methods for characterizing and extracting inspection features are proposed and discussed. A sequential knowledge based filtering method is developed to reduce the number of inspection features typically encountered in metrological inspection planning. Finally, a formalized approach for case representation of relevant inspection domain knowledge using a newly developed parametric-list technological feature graph (PLTFG) is presented.     

基于图论的飞机典型装配工艺数据挖掘方法

为实现飞机装配指令编制的快速化、智能化,提出基于图论的飞机典型装配工艺数据挖掘方法。该方法将停用词表和分隔数组两个约束引入语义分词法完成装配工艺指令的分词处理,并以装配工艺特征向量结构化地表达分词后的装配工艺指令。将装配工艺特征向量映射为单向有序链,以此构建装配工艺图。通过定义装配工艺图的节点支持度和边支持度,给出了典型装配工艺挖掘方法。以某型飞机的平尾装配工艺为例,在多个支持度下验证了方法的有效性。     

A feature-based fixture design methodology for the manufacturing of aircraft structural parts

In fixture design for the manufacturing of aircraft structural parts, there are various challenges and serious problems in industry, such as difficulty in design knowledge capture and reuse, and the arbitrary or non-standard nature of design, which influence the efficiency and quality of fixture design. This paper proposes a feature-based fixture design methodology in which previous fixture design cases and design rules are described in association with features and thus the design knowledge is integrated with geometric information of aircraft structural parts, which are the main concern of this project. In this methodology, machining features of the structural parts and their associated attributes are identified by feature recognition technique from the 3 dimensional (3D) part models defined based on model-based definition (MBD) technique. The feature-based part information models are then established and are used to retrieve previous fixture design cases and design rules stored in the knowledge base. Fixture designers will choose the appropriate retrieved design cases as the starting point for new fixture design or use previous designs to assess his/her current design. In this way the process of new fixture design can be improved. In the current stage of the research, a prototype feature-based fixture design system based on CATIA for the manufacturing of aircraft structural parts has been developed and used by a large airplane manufacturer.     

文本分类中基于概念映射的二次特征降维方法

对高维特征集的降维是文本分类的一个主要问题。在分析现有特征降维方法的基础上,借助《知网》提出一种新的二次降维方法：采用传统的特征选择方法提取一个候选特征集合;利用《知网》对候选集合中的特征项进行概念映射,把大量底层分散的原始特征项替换成少量的高层概念进行第二次特征降维。实验表明,这种方法可以在减少文本语义信息丢失的前提下,有效地降低特征空间维数,提升文本分类的准确度。     

基于区域行进策略的飞机油箱检查机器人路径规划算法

针对类似于飞机油箱环境中连续型机器人的路径规划问题,设计基于区域行进策略的路径规划算法,结合机器人本体结构约束规划到达油箱内任意给定目标点的路径。连续型机器人具有运动灵活性,但超冗余自由度导致了三维空间规划的多解性,增加了算法的复杂度。采用降低维度的方式,通过将三维空间转化为二维平面进行规划,降低了算法的时间复杂度。将飞机油箱的单舱划分为两个区域,根据目标点所处区域位置确定规划策略。最后,基于Matlab对所提算法进行仿真,实验结果验证了算法的可行性和有效性。     

流程制造系统辨识的自动结构性建模方法

模型能够解决系统工程中的许多问题,为此,提出了一种新的流程制造系统辨识的自动结构性建模方法。通过对系统的结构组成和单元关系进行辨识,提炼出模型的结构性数据,并以此自动地形成系统仿真模型。建模采用了图论作为工业系统的数学表达形式。研究对系统单元进行规则性编码,并根据系统结构所具有的特性定义了建模的辨识函数。实例证明了提出的方法是可行的,并能够满足系统建模的有用性、高效性、准确性的要求。     

基于视觉与结构谱特征融合的视频检索

视频的时序结构特征是视频的重要特征。提出了一种基于DTW和图谱理论的视频结构特征提取与表示方法,并将这种结构特征融合视觉特征用于视频的镜头检索。根据镜头分割中的帧差曲线,使用DTW原理得到视频镜头特征帧,以镜头特征帧作为图的顶点,以特征帧间的关系作为图的边,生成镜头关系图,分解图的邻接矩阵,得到镜头结构谱,融合镜头的视觉特征得到最终的检索结果。实验结果表明,视频的结构谱特征可以有效地表示视频的结构特征,在视频检索中也是有效的。     

MC环境下基于加工特征的零件工时定额研究

工时定额是大规模定制生产模式下企业确定产品交货期和提高顾客满意度的重要参考依据。为更好地支持大规模定制生产的企业工时定额的制定,提出一种大规模定制环境下基于加工特征的零件工时定额的制定方法。在此方法中,运用面向对象方法将零件组内的加工特征进行分类、编码,建立加工特征信息模型,以便于零件加工特征的检索;采用神经网络技术,结合MATLAB软件针对编码系统中每一个最底层的加工特征建立对应的工时模型;根据零件加工特征的编码检索各个加工特征的工时模型,估算每个加工特征的加工工时,从而得到整个零件的工时定额。通过与传统的工时定额方法对比,验证了该方法的准确性和快速性。     

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。     

一种基于特征聚类的特征选择方法

特征选择是数据挖掘和机器学习领域中一种常用的数据预处理技术。在无监督学习环境下,定义了一种特征平均相关度的度量方法,并在此基础上提出了一种基于特征聚类的特征选择方法 FSFC。该方法利用聚类算法在不同子空间中搜索簇群,使具有较强依赖关系（存在冗余性）的特征被划分到同一个簇群中,然后从每一个簇群中挑选具有代表性的子集共同构成特征子集,最终达到去除不相关特征和冗余特征的目的。在 UCI 数据集上的实验结果表明,FSFC 方法与几种经典的有监督特征选择方法具有相当的特征约减效果和分类性能。     

Fault detection and diagnosis method for batch process based on ELM-based fault feature phase identification

Because of the multiplicity of operation phases in batch process, which have specific control objects, different dominant process variables and distinct process correlation characteristics, the faults may also have phase characteristic. To conduct fault diagnosis for batch process more precisely, this paper proposes a fault detection and diagnosis method based on fault feature phase identification results. Firstly, extreme learning machine is used to identify fault feature phases between the faulty data set and the normal data set. Then, focusing on the different data nature implied in different fault feature phases, several ‘short stages’ are partitioned for the whole batch. After that, different multiway fisher discriminant analysis (MFDA) models are developed for these ‘short stages,’ respectively. The proposed method can deepen the search space analyzed by fault diagnosis into specific fault feature phases, which not only overcome the disadvantage of too many models in MFDA, but also overcome the disadvantage of low diagnosis accuracy and high false recognition rate of traditional MFDA method. Simulation results show the feasibility and validity of the proposed method.     

基于光照不变感知特征的打火石检测识别

为了在生产过程中确定打火石的方向,提高打火石安装定位的效率,根据打火石圆柱状外形及两端面的颜色特征,使用特定的棋盘形工装定位装置对打火石进行定位,并利用彩色机器视觉技术对打火石的端面颜色进行检测识别。为了准确识别出打火石的品红色端面,在拮抗色感知特征基础上,针对现场环境中光线不均等问题,进行了光照不变的拮抗色感知特征变换,降低了光照不均对检测结果的影响。对于打火石另一端面银白色特征,根据其高亮的特性,采用Otsu自适应阈值法进行分割,并在颜色通道的统计基础上,对银白色特征提取过程进一步进行约束。实验结果表明,采用光照不变的拮抗色感知特征提取算法以及阈值分割相结合的方法,能够有效对打火石的端面颜色进行识别,在不出现误检的情况下,品红色端面的识别准确率为99.87%,银白色端面的识别准确率为99.62%,满足了打火石视觉检测识别系统的性能要求。