数字地图辅助的双星时差频差联合定位方法

针对双星时差、频差联合定位精度受高程误差影响较大问题,提出了一种数字地图辅助的定位方法。在推导出初始定位算法的基础上,引入三维数字地图以降低零高程假设导致的定位误差,并给出了该方法的定位误差的理论表达式,仿真结果表明了该方法的有效性。     

一种人工数字地图的生成算法及其仿真研究

人工数字地图是地形辅助导航系统的重要组成部分.针对实际地形的特点,将人工数字地图的生成过程分成整体骨架的构建与局部细节的调整两个环节.在整体骨架的构建过程中以地形的统计特性为依据,用地形起伏均方差描述某区域地形偏离地形均值的程度,用抗相关距离描述地形点之间相关程度;在局部细节的调整过程中对分形自适应地形生成算法作适当改变,根据地形的斜率信息获取线性化的等效网格距离,并取代随机位移模型.仿真结果与真实地形的统计特点基本吻合,表明该算法能较好地模拟真实的地貌.     

一种改进的地形熵匹配导航算法

提出了一种改进的地形熵辅助导航算法.因为地形熵算法丢掉了地形的相对位置信息及对地形数据做了归一化处理,使得其在地形跟随/地形回避中的应用存在着匹配误差容易发散的问题.基于此点,提出了基于卡尔曼滤波的惯性地形辅助导航方法,交互信息量等方法的地形熵算法,以改进该算法在地形跟随/地形回避中应用的性能.最后用matlab软件根据分形学中Weierstrass-Mandelbrot函数产生了模拟的数字地图,并对该地形匹配导航方法作了仿真研究,仿真结果表明该方法可以抑制误差的发散,并提高了地形匹配的定位精度.     

用约束四叉树实现地形的实时多分辨率绘制

在地形可视化领域，实时绘制复杂地形场景的最有效工具是LOD技术，在研究和实验的基础上，提出了一种基于地形四叉树实时构建地形多分辨率模型的优化算法，该算法引入视相关的概念，给出一种与视点相依赖的对地形结点误差进行评价的方法，改进了LOD模型“裂缝”效应消除方法。实验结果表明，该算法能实时动态地生成地形的连续多分辨率模型，实现地形场景的平滑绘制。     

大面积人工数字地图生成软件的设计与实现

为给地形匹配研究提供低成本、大面积、生成速度快的随机数字地图,开发了一种数字地图生成软件.选用二维离散AR算法产生模拟数字地形,采用小块地图拼接的方法生成大面积地图.用标准C语言编程实现地形数据的快速生成和调用,并根据要求可以任意加入威胁地形.建立了便于调用数据的地形数据存储格式.利用该软件生成大面积数字地图速度快,面积可以按需要设定,生成的地形符合实际地形特征.由于每次生成的地形特征可以不相同,所以该软件可为地形匹配研究提供有效的大面积地形数据.     

无人机航路规划环境模型研究

通过威胁信息的分层转化,利用信息融合原理构造无人机航路规划的环境模型。在对原始数字地图二维三次插值处理的基础上,分别针对无人机的最大爬升坡度、最小离地间隙以及纵向转弯曲率约束进行了数字地形的调整。分别对探测威胁和火力威胁进行分析,以雷达发现概率、水平距离以及遮蔽影响因素构造探测威胁转化模型,以击落概率和导弹最大作用半径构造火力威胁转换模型,以地形因素的形式融入预处理后的数字地图中,经相关约束调整构造出无人机航路规划环境模型。     

改进BP神经网络模式识别在消除惯导漂移误差中的应用

为了能实时地消除惯导陀螺漂移误差,在数字地图为基础的前提下,引入了改进BP神经网络模式识别技术,并设计了相应的三维数字地形辅助惯性组合导航算法——BP算法。在匹配过程中,分别给定了不同的数字地形模式和不同载体的机动轨迹。利用C＋＋程序进行算法中的数值计算,然后通过Matlab软件模拟导航数值仿真。多次实验证明,提出的方法具有很好的可靠性和可行性,具有重要的工程应用价值和理论借鉴意义。     

寒冷干旱区土壤水力特性参数的模拟估算

测量非饱和带水力特性参数，诸如：饱和含水率、田间持水率等系数，即耗时、费力又难以保证试验精度。但利用实测的土壤基本物化特征参数，颗粒级配、土壤容重等，通过土壤传递函数(PTFs)可以较准确地内插估算出土壤水力特性参数。本研究利用国家自然基金重点项目区——内蒙古太仆寺旗试验断面的实测资料，引入对数平均粒径和对数标准偏差，通过人工神经网络方法建立了太仆寺旗研究区的BP-PTFs函数模型和BP-PTFs-log函数模型，用误差分析理论分析评价了两函数模型的有效性、适用性和精度，并通过两种函数模型的外推预测结果的分析检验比较，结果表明，所建立的BP-PTFs-log函数模型要优于BP-PTFs模型。     

考虑地形影响的卫星导航性能评估方法

研究准确评估导航性能,针对地形因素会影响导航卫星的可见性,为了准确评估地形复杂地区对卫星导航的性能的影响,在仿真模型中引入地形影响因素.利用可视域分析理论完善了卫星可见性的判断依据,给出了考虑地形影响时可见卫星颗数和精度因子的分析模型,做为指标对卫星导航性能进行评估.结合机场终端区数字高程模型、卫星星历、航图等数据,对...     

空间自适应正则化超分辨率图像重建

超分辨率图像重建是一个病态问题,在重建过程中需要正则化处理,而正则化重建会引入正则化误差及重建过程中由于病态性而引入的噪声放大误差,且这两类误差均和图像的空间局部特性有关.提出根据图像的局部空间统计特性自适应控制超分辨率图像正则化重建算法,采用图像局部统计方差来区分图像棱边区域及平滑区域,在图像的棱边区域加强图像的约束重建,而在图像的平滑区域加强正则化.实验表明该算法能有效地减小重建误差,算法的信噪比得益优于传统的正则化重建算法及总变分模型重建算法,并且对正则化参数的选择具有一定的鲁棒性.     

双目立体视觉系统的精度分析

为提高双目立体视觉系统的测量精度,分析双目视觉系统的工作原理以及该系统各个环节的误差模型.通过提高标定模板的精度、增加标定图像,可以减小标定误差及图像校正行对齐误差.分析图像识别误差对视觉系统测量精度的影响,导入视觉系统的结构参数与测量误差之间的关系,从而确定结构参数对测量精度的作用.讨论物距对测量精度的影响,并给出合...     

用于非高斯系统降维的最小残差熵主元网络

基于最小均方误差的主元分析和主元神经网络是有效的多变量降维统计技术,它们所提取的主元含有系统最大方差.非高斯随机系统的近似模型应当含有系统最大信息熵,但包含最大方差并不一定包含最大信息熵.该文提出一种以最小残差熵为通用指标的非线性主元神经网络模型,并给出了一种基于Parzen窗口密度函数估计的熵近似计算方法和网络学习算法.然后从信息论角度分析了,在高斯随机系统中基于最小残差熵和最小均方差为指标的主元网络学习结果具有一致性.最后以仿真验证该方法的有效性,并与基于最小均方误差的主元分析和主元神经网络方法的计算结果进行对比性分析.     

Inherited error in finite element analyses of structures

Mathematicians are quick to point out that round-off and truncation errors induced by the digital computer are only half of the manipulation errors in numerical analyses. The other half are the errors in quantizing the mathematical problem for computer solution: errors inherited for the equation solving process.This paper examines the relevance of inherited errors in structural analyses using the finite element concept and the digital computer. It illustrates error magnitudes using numerical experiments of simple structures. It constructs a theory explaining the errors for these systems. Having identified the most significant controllable computer error source, it describes a process for minimizing its contribution to the inherited error.The paper concludes that orders of magnitude between errors reported by various investigators can be explained by differences in inherited error. The most significant effect of these errors can be identified with inconsistency in problem formulation. These inconsistencies can be eliminated by exploiting the existence of rigid body states in the finite element models. Thereby, solution errors introduced by inherited errors can be reduced to intrinsic errors in parameters defining the geometry, material characteristics, and boundary conditions.     

A software cost model with imperfect debugging,random life cycle and penalty cost

The paper develops a cost model with an imperfect debugging and random life cycle as well as a penalty cost that is used to determine the optimal release policies for a software system. The software reliability model, based on the nonhomogeneous Poisson process, allows for three different error types: critical, major and minor errors. The model also allows for the introduction of any of these errors during the removal of an error. Using the software reliability model presented, the cost model with multiple error types and imperfect debugging is developed. This cost also considers the penalty cost due to delay for a scheduled delivery time and the length of the software life cycle is random with a known distribution. The optimal software release policies that minimize the expected software system costs (subject to the various constraints) or maximize the software reliability subject to a cost constraint, are then determined. Numerical examples are provided to illustrate the results.     

Robot measuring form errors

This work presents an automated system for the measurement of form errors of mechanical components using an industrial robot. A three-probe error separation technique was employed to allow decoupling between the measured form error and errors introduced by the robotic system. A mathematical model of the measuring system was developed to provide inspection results by means of the solution of a system of linear equations. A new self-calibration procedure, which employs redundant data from several runs, minimizes the influence of probes zero-adjustment on the final result. Experimental tests applied to the measurement of straightness errors of mechanical components were accomplished and demonstrated the effectiveness of the employed methodology.     

Influence of sensor imperfections to electronic compass attitude accuracy

Analysis of various errors of orientation and motion sensors and their influence on resulting compass attitude information are investigated and quantified in this paper. A triple-axis acceleration sensor and a triple-axis magnetic sensor are considered to form a compass module giving full orientation information (attitude). 3D-compass module with PCB-fluxgate sensors and off-shelf acceleration sensor is introduced. Three main sensor error sources are discussed separately and their contribution to final compass heading accuracy is investigated. These are sensor orthogonality error, linearity error and ADC quantization noise. The influence of these errors is simulated on artificial group of test cases which evenly covers various orientations of the compass module. It was found that the errors in the accelerometric system have major influence on the heading accuracy. It was concluded that even low-resolution AD converters may have only a minor influence on the system accuracy, while the dominant error sources are sensor triplet nonorthogonality and sensor non-linearity.     

多新息抗差-自适应卡尔曼滤波定位算法研究

针对移动机器人在定位过程中,由传感器测量误差和机器人模型引起的位姿误差导致系统定位精度急剧下降的问题,提出了一种多新息卡尔曼滤波算法.在标准卡尔曼滤波的基础上,当传感器测量值存在误差时,引入抗差权因子,通过改变误差测量值的权值提高滤波器的估计精度;当机器人位姿存在误差时,引入自适应因子,通过调整状态协方差矩阵的大小抵制位姿误差引起的滤波发散.同时,引入了多新息,即多个时刻的新息向量,进一步提高此非线性系统的精度.实验表明:当存在测量误差和位姿误差时,该滤波算法能有效提高定位精度.     

Error-driven active learning in growing radial basis function networks for early robot learning

In this paper, we describe a new error-driven active learning approach to self-growing radial basis function networks for early robot learning. There are several mappings that need to be set up for an autonomous robot system for sensorimotor coordination and transformation of sensory information from one modality to another, and these mappings are usually highly nonlinear. Traditional passive learning approaches usually cause both large mapping errors and nonuniform mapping error distribution compared to active learning. A hierarchical clustering technique is introduced to group large mapping errors and these error clusters drive the system to actively explore details of these clusters. Higher level local growing radial basis function subnetworks are used to approximate the residual errors from previous mapping levels. Plastic radial basis function networks construct the substrate of the learning system and a simplified node-decoupled extended Kalman filter algorithm is presented to train these radial basis function networks. Experimental results are given to compare the performance among active learning with hierarchical adaptive RBF networks, passive learning with adaptive RBF networks and hierarchical mixtures of experts, as well as their robustness under noise conditions.     

语音识别错误的分类分析

大词表连续语音识别系统由多个组件构成,识别错误受多种因素的影响。系统开发者需要分析错误发生的不同原因。根据语音识别的基本理论给出了对错误进行分类分析的原理,将识别错误按错误原因分为解码错误、声学模型错误、语言模型错误、声学和语言复合错误四大类,并对分类后的错误做了统计分析。实验证明,识别错误的分类分析为系统的改进提供了参考依据。     

超大范围坐标测量系统的现场测试基准及其误差分析

CCD立体视觉坐标测量系统已有广泛应用,但其有效测量范围通常在十数米之内.超大范围坐标测量系统是一种适于野外环境的坐标测量系统,其测量范围可覆盖直径达数百米的圆形区域.文章从分析测量系统数学模型入手,指出了建立现场测试基准的必要性,随后给出一种既能保障测量结果正确可靠又可简化测量模型的基准建立方法,并分析了基准建立过程中各引入误差对系统误差的影响.实验数据证明了文中测量原理及测试基准建立方法的可行性.