基于粒子群优化的理论变异函数拟合方法研究

变异函数是地统计学中区域化变量空间结构分析和空间局部插值的主要分析工具.理论变异函数模型的获取是地质统计学中的基础性工作,它是了解区域化变量的变异特征、进一步对地质统计学计算的必要环节.针对现有的理论变异函数的拟合方法,如人工拟合法、线性规划拟合法、加权多项式拟合法、目标规划拟合法等的不足之处,充分利用粒子群优化算法在求解非线性优化问题时具有的全局寻优的特点,提出基于粒子群优化的理论变异函数拟合方法.在实例应用中,分别利用粒子群优化算法和加权多项式拟合方法进行理论变异函数拟合,交叉验证结果表明粒子群优化算法预测精度较高,具有较强的稳健性.     

SAR波长对遥感图像统计特征的影响

遥感图像统计特征在目标辨识、图像配准以及图像分类等任务中扮演着重要角色。传统的统计特征鲁棒性分析一般是基于图像具有同一传感器参数的假设。而在现实任务中,往往需要处理由不同传感器参数拍摄的遥感图像,比如不同波长或入射角度等。传感器参数不同会导致图像灰度与表象特征不同,进一步会影响到统计特征表达。因此有必要系统研究常用统计特征与传感器参数之间的相关性,从而实现更加鲁棒的图像处理与分析。本文研究了目前常用的六种图像统计特征,在合成孔径雷达（SAR）波长变化时（L波段、C波段和X波段）的鲁棒性。实验中,对鲁棒性能指标作出定义,并对不同统计特征的鲁棒性做出了比较与分析。实验结果发现,不同统计特征对波长变化的鲁棒性不同,同时图像特征在不同地形时随波长变化的规律也不一致。本文的研究成果对基于SAR图像的判读、景象匹配制导以及景象适配性分析等具有重要的指导意义。     

利用光谱量值和形状优化组合的地表覆盖变化检测方法

分析了几种常见的变化检测算法的鲁棒性,在此基础上提出一种光谱量值和形状优化组合的地表覆盖变化检测方法。以西安市及其周边为研究区域,分析该区域2000～2009年的地表覆盖变化。变化检测的总体精度为92.313%,Kappa系数为0.844,优于其他传统算法。     

Real time hybrid simulation: from dynamic system,motion control to experimental error

Real‐time hybrid simulation (RTHS) has increasingly been recognized as a powerful methodology to evaluate structural components and systems under realistic operating conditions. It is a cost effective approach compared with large scale shake table testing. Furthermore, it can maximally preserve rate dependency and nonlinear characteristics of physically tested (non)structural components. Although conceptually very attractive, challenges do exist that require comprehensive validation before RTHS should be employed to assess complicated physical phenomena. One of the most important issues that governs the stability and accuracy of an RTHS is the ability to achieve synchronization of boundary conditions between the computational and physical substructures. The objective of this study is to propose and validate an H_∞ loop shaping design for actuator motion control in RTHS. Controller performance is evaluated in the laboratory using a worst‐case substructure proportioning scheme. A modular, one‐bay, one‐story steel moment resisting frame specimen is tested experimentally. Its deformation is kept within the linear range for ready comparison with the reference closed‐form solution. Both system analysis and experimental results show that the proposed H_∞ strategy can significantly improve both the stability limit and test accuracy compared with several existing strategies. Another key feature of the proposed strategy is its robust performance in terms of unmodeled dynamics and uncertainties, which inevitably exist in any physical system. This feature is essential to enhance test quality for specimens with nonlinear dynamic behavior, thus ensuring the validity of the proposed approach for more complex RTHS implementations. Copyright © 2012 John Wiley & Sons, Ltd.     

洪水频率分析稳健性方法研究

结合使用Huber的M估计和最小距离(MD)估计,构造了一种P-Ⅲ型总体下的频率分析方法。统计试验结果表明,当样本中有一定比例干扰数据的影响时,该方法具有较好的稳健性;同时,当样本中无干扰数据的影响时,该方法亦具有相对较优的统计性能。     

基于小波变换的栅格数字地图水印算法

本文提出了一种基于小波变换的栅格数字地图水印算法。算法基于栅格数字地图相比自然图像具有较高亮度,低频分量具有更高感觉容量的特点,在栅格数字地图经小波变换后的低频分量部分依据其亮度调节强度嵌入周期性扩展水印;同时为提高算法鲁棒性,在细节分量部分嵌入相同水印,水印嵌入位置根据人类视觉系统特性和小波系数间内在联系确定。实验分析表明,该算法能较好抵抗诸如JPEG压缩、锐化、剪裁等各种攻击,具有好的鲁棒性。     

一种矢量线状要素数据零水印算法

提出一种针对矢量线状要素数据的零水印算法。首先,利用Arnold技术置乱原始水印图像;然后,采用Douglas-Peucker压缩算法提取矢量线状要素数据的特征点;最后,将置乱后的水印图像与由相邻特征点所构成直线的斜率调制所得的二值矩阵进行异或运算,从而构造出零水印图像。实验结果表明,该算法能够抵抗常见的平移、旋转、缩放等几何攻击,并且对Douglas-Peucker压缩和裁剪攻击具有较好的鲁棒性。     

子波估计及极大似然反褶积方法稳健性讨论

利用极大似然反褶积提取鲁棒性地震子波的方法进行了研究，并讨论了极大似然反褶积的稳健性。理论分析表明，在理想条件下极大似然反褶积是无偏估计，而且还是一致估计，这一结论可以由所提取的地震子波得到验证。实际地震资料的处理结果分析表明，极大似然反褶积具有很好的稳健性。应当指出，为保证地震子波的鲁棒性，极大似然反褶积子渡时窗应尽量满足统计条件，以提高极大似然反褶积的稳健性。     

Modified model-based control of shake tables for online acceleration tracking

The interest in shake tables stems from a need to simulate earthquake behavior in laboratory settings. However, the inherent properties and nonlinearities associated with electromechanical and servohydraulic shake tables, combined with issues of table-structure interaction, make accurate reproduction of earthquake acceleration time histories a challenging problem. The classical approach to control shake tables has been the Transfer Function Iteration (TFI) method. The tuning of the TFI controller is an offline iterative process, conducted using small amplitude ground motions. Effective compensation is not achievable for system nonlinearities that are not projected in the iterative tuning process. To address this problem, researchers have developed online compensation techniques, which can maintain tracking performance for the earthquake signals more effectively. Model-based controllers (MBC) are a class of online controllers which use an identified model of the shake table-structure for compensation. The MBC employs feedforward and feedback controllers to ensure that the shake table tracks a specified earthquake ground motion despite the presence of table and structural nonlinearities. However, the feedback controllers in MBC do not always maintain tracking accuracy and can result in loss of robustness when changes occur in the shake table and structure dynamics. This paper introduces a modified model-based controller (mMBC) for acceleration tracking as an improvement on the existing MBC architecture. A stability condition is introduced to assess the robustness of the new modified control architecture. Through numerical and experimental studies, the improved tracking robustness of the mMBC architecture is demonstrated.     

基于小波变换的矢量地理空间数据数字水印算法

基于矢量地理空间数据自身特点,运用小波变换技术,提出了一种基于小波变换的矢量地理空间数据数字水印算法.即对一系列的线、面坐标串分别进行离散小波分解,把水印信息分段嵌入到相应的小波分解后得到的低频系数中;再进行小波逆变换得到含水印信息的矢量地理空间数据.在水印信息提取过程中,解决了删点、投影变换等对水印信息提取的不利影响.实验分析表明,该算法在抗矢量地理空间数据处理中常见的噪声攻击、坐标数据轻微的移动等方面具有较好的效果.