范例推理在口腔创伤诊断系统中的应用

鉴于传统口腔创伤诊断专家系统在知识获取上存在的瓶颈问题,提出一种基于范例推理(CBR)的口腔创伤诊断系统.首先,针对传统口腔创伤诊断的思维推理过程,给出了口腔创伤诊断问题中各要素的系统化定义,建立系统的的概念模型.然后,具体讨论了口腔创伤诊断系统中范例库和结果库的建立,并将其用关系数据库的格式存储.最后,应用模糊集和灰色关联理论,给出了关联度的定义和计算方法,从而能够准确地检索到相似诊断范例,进而对口腔创伤诊断给出定性或定量的分析结果.     

面向结构化场景的激光雷达点云高精度配准与定位方法

在基于先验地图的激光雷达室内导航方案中,通常采用点云配准的方法进行无人设备位姿初始化.在结构化场景下,传统配准算法特征鲁棒性较差,导致点云配准的误差较大且易陷入局部最优.针对该问题,提出了一种基于多平面空间模型的点云快速配准方法.首先该方法利用特征直方图的思想对空间点云进行快速粗聚类,根据平面一致性将粗聚类后的点集进行...     

固体火箭发动机体空间缺陷精确定位方法

在固体火箭发动机体数据三维重构的基础上,构建出符合脱黏、气泡和夹渣体数据特征的三维相似变换模板,经三维相似变换后,定位出各类缺陷的体数据场位置,选取缺陷几何中心为缺陷精确定位的标志点,给出了体空间中缺陷几何中心的计算方法.设计制作了预置典型缺陷的模拟固体火箭发动机,测量得到的缺陷几何中心和实际几何中心的距离与最大直径的比值都不大于10%,即测量得到的几何中心与实际几何中心很接近,因此对缺陷几何中心的定位是准确的.      

Geometric error analysis of an over-constrained parallel tracking mechanism using the screw theory

This paper deals with geometric error modeling and sensitivity analysis of an overconstrained parallel tracking mechanism. The main contribution is the consideration of overconstrained features that are usually ignored in previous research. The reciprocal property between a motion and a force is applied to tackle this problem in the framework of the screw theory. First of all, a nominal kinematic model of the parallel tracking mechanism is formulated. On this basis, the actual twist of the moving platform is computed through the superposition of the joint twist and geometric errors. The actuation and constrained wrenches of each limb are applied to exclude the joint displacement. After eliminating repeated errors brought by the multiplication of wrenches, a geometric error model of the parallel tracking mechanism is built. Furthermore,two sensitivity indices are defined to select essential geometric errors for future kinematic calibration. Finally, the geometric error model with minimum geometric errors is verified by simulation with SolidWorks software. Two typical poses of the parallel tracking mechanism are selected, and the differences between simulation and calculation results are very small. The results confirm the correctness and accuracy of the geometric error modeling method for over-constrained parallel mechanisms.     

基于局部几何特征的稠密点云配准方法

针对现有的稠密点云配准方法依赖初始位置设定、计算成本高、配准成功率不高等问题，提出了一种基于点云局部几何特征的稠密点云配准方法。采用深度卷积网络模型提取点云的局部几何特征，从而减少了三维点云数据的噪声、低分辨率和不完备性等带来的影响。在此基础上，使用K维树搜索完成局部几何特征描述子的关联工作。最后，通过随机采样一致算法对点云的相对位姿进行鲁棒的估计。通过对开源数据集上5个典型场景中的数据测试表明，该方法的配准成功率达到92.5%，配准精度达到0.0434m，配准时间相对最邻点迭代配准算法缩短了74.7%，实验结果验证了该方法的有效性、实时性和鲁棒性。     

基于遗传算法和 Vague集的液压位置控制系统自适应 PID控制

在被控对象模型精度不高的情况下,设计了一种自适应PID控制方法,提高其鲁棒性,以确保系统的响应具有最优的动态性能和稳态性能。该方法在常规PID控制的基础上,引入智能技术,在线实时调整PID控制器的3个参数。该控制器主要由3部分构成:一是采用遗传算法优化模糊推理规则;二是精确的Vague集推理规则表;三是基于Vague集相似度量的自适应PID控制。仿真结果表明,该控制方法响应速度快,稳态精度高。     

分布式IMA的网络分区方法及其实时性能分析

 针对分布式综合模块化航空电子(IMA)体系结构,在时间、空间分区以及带宽分区的基础上,提出了网络分区的概念,并建立了相应的网络模型、消息模型、流量模型及调度模型。利用网络演算方法,推导了航空电子混合消息集在网络分区下端到端(ETE)延迟的计算公式;搭建典型网络,通过理论计算对比了网络分区与带宽分区的实时性能。计算结果表明:网络分区下两条硬实时数据流的延迟较带宽分区分别降低了33.5%和74.2%;而弱硬实时与软实时数据流的延迟增加约30%。最后通过OPNET仿真对理论分析所得结果进行了验证,证明网络分区方法符合分布式IMA的分布式架构要求,并满足了系统混合关键性的需求。     

基于光照模糊相似融合不变矩的航天器目标识别

针对航天器目标识别中现有的光照变化和图像模糊条件下几何仿射不变矩识别不稳健问题,根据灰度光照模型,结合图像模糊和不变矩的基本理论,提出了一种光照模糊不变量,与几何仿射不变矩进行融合,构建出一种新的光照模糊相似融合（CIBA）不变矩,并从理论上证明了光照、模糊和相似变换（平移、缩放和旋转）不变性。对3种不同的航天器目标的识别实验表明,光照模糊相似融合不变矩在最小距离分类器中的识别准确率达到了86.37%,比几何仿射不变矩提高了94.00%,比具有光照不变性的几何仿射不变矩提高了24.54%,有效地解决了对不同位姿、不同光照模糊条件下的航天器目标识别问题,提高了基于矩的目标识别鲁棒性。     

复杂结构件数控编程加工特征用户自定义方法

加工特征是实现复杂结构件高效、高质量数控编程的有效手段,但是同一类加工特征只是几何形状和加工工艺相似,并不完全相同。如何适应不同的企业资源与工艺水平、不同类型的复杂结构件进行加工特征定义是基于加工特征进行自动数控编程的一个难题。针对以上难题,本文提出了一种复杂结构件数控编程加工特征用户自定义方法,基于全息属性面边图表达加工特征几何信息,给出了具有一定柔性的加工特征几何信息定义方法,基于语义与规则建立加工特征工艺信息及其与几何信息之间的关联关系,实现了由用户根据企业的制造资源、零件结构和工艺人员的编程习惯等因素自定义加工特征。根据本文提出的方法开发了飞机复杂结构件加工特征用户自定义及自动编程系统,已成功应用于国内某大型航空制造企业的飞机结构件数控编程,经过多项飞机结构件测试,本文提出的方法特征识别正确率平均达到97%。     

基于OSPA距离的航迹关联方法

 提出一种基于最优次模式分配(OSPA)距离的多传感器航迹关联方法,把每一个局部传感器航迹看做一个航迹集合,通过OSPA距离度量各个航迹集合之间的距离,具有最短OSPA距离的航迹属于同一个目标航迹;针对航迹关联实时性的要求,进一步给出具有滑窗的航迹OSPA递推方法。最终结果表明,本文方法不仅可以有效保持航迹的历史信息和几何特征信息,而且可以有效地关联航迹交叉、航迹分叉和航迹异步等问题。通过与加权关联方法、独立序贯方法的仿真比较,证明其在关联精度、异步等几个方面具有比较好的优势。     

考虑几何分散性的涡轮盘寿命概率分析

针对考虑几何分散性的涡轮盘低循环疲劳（LCF）寿命概率分析中几何参数多、几何随机变量难确定、分布特征获取困难、模型需自动更新及计算成本高的问题,提出几何分散性的概率处理方法:采用试验设计方法对涡轮盘结构所有几何参数进行灵敏度分析,筛选出对应力影响较大的关键几何参数作为随机变量,使用K-S(Kolmogorov Smirnov)方法确定其分布类型和特征参数,最后建立代理模型进行Monte Carlo概率分析.基于此方法,开发出了涡轮盘概率分析系统,在该系统中筛选得到某发动机GH720Li涡轮盘内径、外径、盘缘厚度3个结构参数作为几何随机变量,完成对LCF寿命的概率分析工作得到寿命-可靠度分布曲线.分析结果表明涡轮盘外径对LCF寿命有较大影响.      

Hybrid discriminative/class-specific classifiers for narrowband signals

The class-specific (CS) method of signal classification operates by computing low-dimensional feature sets defined for each signal class of interest. By computing separate feature sets tailored to each class, i.e., CS features, the CS method avoids estimating probability distributions in a high-dimension feature space common to all classes. Building a CS classifier amounts to designing feature extraction modules for each class of interest. In this paper we present the design of three CS modules used to form a CS classifier for narrowband signals of finite duration. A general module for narrowband signals based on a narrowband tracker is described. The only assumptions this module makes regarding the time evolution of the signal spectrum are: (1) one or more narrowband lines are present, and (2) the lines wandered either not at all, e.g., CW signal, or with a purpose, e.g., swept FM signal. The other two modules are suited for specific classes of waveforms and assume some a priori knowledge of the signal is available from training data. For in situ training, the tracker-based module can be used to detect as yet unobserved waveforms and classify them into general categories, for example short CW, long CW, fast FM, slow FM, etc. Waveform-specific class-models can then be designed using these waveforms for training. Classification results are presented comparing the performance of a probabilistic conventional classifier with that of a CS classifier built from general modules and a CS classifier built from waveform-specific modules. Results are also presented for hybrid discriminative/generative versions of the classifiers to illustrate the performance gains attainable in using a hybrid over a generative classifier alone.     

信息融合图像识别算法及其在三维飞机图像识别中的应用研究

将D S证据推理与不变矩理论相结合,提出了信息融合图像识别算法,并应用于三维飞机图像的识别。构造了基本置信指派函数,建立了分类规则,针对Dempster组合规则只适用于高置信度、低冲突性的情况,应用可解决证据冲突的加权分配冲突法及吸收法,完成了三维飞机图像识别的仿真,比较了3种组合规则的识别效果,结果表明吸收法识别率最高。     

跑道侵入严重度本体建模及等级划分研究

为研究致险因素对跑道侵入严重度的影响,从知识建模角度出发,构建了跑道侵入严重度划分的领域本体模型.采用石川图针对跑道侵入严重度划分问题进行因果分析;进而以跑道场景要素为导向构建本体模型;应用FaCT++推理机对实际案例分析推理,结合DL query 模块辅助验证推理结果.结果表明,本体推理可以支持跑道侵入严重度划分,验证了模型的有效性.     

Solar Weather Event Modelling and Prediction

Key drivers of solar weather and mid-term solar weather are reviewed by considering a selection of relevant physics- and statistics-based scientific models as well as a selection of related prediction models, in order to provide an updated operational scenario for space weather applications. The characteristics and outcomes of the considered scientific and prediction models indicate that they only partially cope with the complex nature of solar activity for the lack of a detailed knowledge of the underlying physics. This is indicated by the fact that, on one hand, scientific models based on chaos theory and non-linear dynamics reproduce better the observed features, and, on the other hand, that prediction models based on statistics and artificial neural networks perform better. To date, the solar weather prediction success at most time and spatial scales is far from being satisfactory, but the forthcoming ground- and space-based high-resolution observations can add fundamental tiles to the modelling and predicting frameworks as well as the application of advanced mathematical approaches in the analysis of diachronic solar observations, that are a must to provide comprehensive and homogeneous data sets.     

AFIT neural network research

A brief summary of research done at the Air Force Institute of Technology (AFIT) in the area of neural networks is provided. It has been shown that backpropagation, used for feedforward artificial neural networks, is just a degenerate version of an extended Kalman filter, and that networks can do about as well as the optimum statistical classification technique. A method of finding the importance of features for use by a neural network classifier has been determined. Techniques for using neural networks for image segmentation have been developed. In optical pattern recognition, techniques that allow the processing of real FLIR (forward-looking infrared) images with existing binary spatial light modulators have been devised. An optical direction of arrival detector applicable to laser illumination direction determination has been designed and tested; the design is similar to a fly's eye. Coated mirrors for the optical confocal Fabry-Perot interferometer have been designed, specified, fabricated, and installed. Significant progress has been made in the use of neural networks for processing multiple-feature sets for speech recognition     

混合式壁面移动机器人横向移动机构及运动分析

 针对结构较复杂的椭球形壁面,提出了由两类不同移动机构组成的混合式壁面移动机器人,一类为框架移动式,实现沿壁面经线（纵向）的攀爬运动;另一类为浮动的轮轨驱动式,实现沿壁面纬线（横向）的运动,二者相互独立。重点介绍了轮轨式横向移动机构,分析了它的工作原理,根据工作环境的几何特征并结合DH法,对复杂约束环境下,机器人的横向运动进行了运动学分析,并进行了样机实验,结果表明通过控制两个驱动轮的角速度可以控制机器人横向运动的速度和姿态。     

考虑几何特征与载荷敏感性的航空发动机结构件疲劳寿命预测模型

为了分析几何特征和载荷因素对几何复杂结构疲劳寿命预测的影响,给出了结构损伤区的定义方法,通过引入几何特征因子和疲劳权函数表征几何特征对疲劳寿命的影响,定义了载荷敏感性因数表征载荷因素对疲劳寿命的影响,以临界距离理论为基础,结合有限元分析、Origin函数拟合及Matlab程序等多种工具,发展了一种考虑结构几何特征与载荷敏感性的缺口疲劳寿命预测模型,并将该模型用于TC4、GH901和DZ125合金的缺口试样低循环疲劳寿命预测,结果表明:对于不同几何特征、不同载荷大小和不同材料等多种情况,寿命的预测值与试验值基本在2倍分散带之内,预测精度较为理想且高于多种现有其他方法,验证了该预测模型具有可行性和优越性。     

A heuristic cabin-type component alignment method based on multi-source data fusion

In cabin-type component alignment, digital measurement technology is usually adopted to provide guidance for assembly. Depending on the system of measurement, the alignment process can be divided into measurement-assisted assembly (MAA) and force-driven assembly. In MAA, relative pose between components is directly measured to guide assembly, while in force-driven assembly, only contact state can be recognized according to measured six-dimensional force and torque (6D F/T) and the process is completed based on preset assembly strategy. Aiming to improve the efficiency of force-driven cabin-type component alignment, this paper proposed a heuristic alignment method based on multi-source data fusion. In this method, measured 6D F/T, pose data and geometric information of components are fused to calculate the relative pose between components and guide the movement of pose adjustment platform. Among these data types, pose data and measured 6D F/T are combined as data set. To collect the data sets needed for data fusion, dynamic gravity compensation method and hybrid motion control method are designed. Then the relative pose calculation method is elaborated, which transforms collected data sets into discrete geometric elements and calculates the relative poses based on the geometric information of components. Finally, experiments are conducted in simulation environment and the results show that the proposed alignment method is feasible and effective.