基于资源限制性人工免疫系统的多光谱遥感影像分类方法

提出了一种基于资源限制性人工免疫系统（resouree limited artificial immune systems, RLAIS ）的多光谱遥感影像分类方法。该方法采用RLAIS对遥感影像分类中选取的感兴趣样区进行样本训练，得到全局聚类中心，利用聚类中心对遥感影像进行分类。实验证明，该方法在分类精度上优于传统方法，其总精度和Kappa系数分别达到了91%和0．88，具有实用价值。     

基于遗传算法最短路径问题的探讨

对用遗传算法求解最短路径问题作了有益的尝试，详细分析了求解最佳路径的遗传算法的构成要素，提出了一种新的交叉算子，并且论证了算法参数对结果的影响。通过仿真实验，给出了算法的主要性能参数，证明了算法的可行性。并指出了遗传算法求解最短路径问题的不足之处。     

基于遗传算法的边坡稳定性评价的动态聚类法

针对常规动态聚类方法对初始聚类中心的敏感性以及聚类结果与样本输入次序有关等问题，提出了基于遗传算法的边坡稳定性评价的动态聚类方法，此方法对三峡库岸36个边坡的研究结果表明，该方法是一个具有全局最优解的动态聚类方法，其结果明显好于常规动态聚类方法。     

A disjoint algorithm for seismic reliability analysis of lifeline networks

The algorithm is based on constructing a disjoin kg t set of the minimal paths in a network system. In this paper, cubic notation was used to describe the logic function of a network in a well-balanced state, and then the sharp-product operation was used to construct the disjoint minimal path set of the network. A computer program has been developed, and when combined with decomposition technology, the reliability of a general lifeline network can be effectively and automatically calculated. Supported by: Key Project of Science and Technology from the State Plan Committee, No. 101-9914003     

空间逐步寻优的数据挖掘法的多波段影像分类研究

对于在特征空间中寻找特征模式,一般是通过假设分布函数一次性对样本空间进行分离的方法去试图获得特征空间的样本总体分布规律。但是由于样本集之间互相重叠或者由于离散样本相互干扰的原因,往往很难获取细节性和过程性的分布结构,而直接影响结果的精度和解释力。本文提出了空间逐步寻优的数据挖掘方法(SOMM),是在遗传算法寻优理论基础上,根据知识参数化样本分布函数,来逐步分离样本空间,获得样本空间的树状的层状分布结构;同时提出了基于SOMM的多波段遥感影像聚类模型和监督分类模型;最后对分类过程和结果进行了综合分析,通过与最大似然和BP神经网络方法相比较,认为SOMM方法在过程化、细节化、分类精度、融合领域知识等方面具有一定的优势。     

MEASUREMENT OF WATER VAPOR AND CLOUD LIQUID WATER CONTENT BY A DUAL-CHANNEL GROUND-BASED MICROWAVE RADIOMETER

The design principle and structural features of a new type dual-channel ground-based microwave radiometerfor measurement of water vapor and liquid water in troposphere are discussed. A method independent ofradiosonde data is used for system absolute calibrations. A statistical method for retrieving the total watervapor and cloud liquid water content and a nonlinear iterative algorithm for retrieving the water vapor profilesin the troposphere are discussed. Comparison is conducted between the measurement of the integrated amountsof the water vapor and the tropospheric vapor profiles by the radiometer and the radiosonde data.     

Rapid solution of kriging equations,using a banded Gauss elimination algorithm

Summary A crucial concern when implementing computer algorithms for geostatistical analyses on microcomputers is speed of execution. Kriging, in particular, typically relies on a Gauss elimination algorithm to solve for weights. Because such an alogrithm is required for each estimate, the solution for weights can result in very slow program execution speed on a microcomputer. One approach to enhancing the efficiency of Gauss elimination is demonstrated herein. The upper triangle plus diagonal of the intersample covariance matrix is used in a modified banded Gauss elimination algorithm. Results show that such an algorithm yields approximately a two-fold reduction in execution time for kriging when the number of nearest neighbours used for estimation is large.     

跨孔地震层析成像的级联方法

对厚度小于1/4波长的超薄波速干扰体进行高分辨率成像,作者曾采用走时反演和波场反演相结合的方法,体现了由低分辨率向高分辨率逐步逼近的思想.级联算法是这种思想的进一步发展,我们将具有不同分辨率的算法串联起来,以达到高分辨率成像的目的.本文介绍一种三级串联的算法,并进行了算法分析,数值计算的例子说明这种级联算法分辨率高、稳定性好,只需要地震资料而不要求其它先验信息,因此能较好地满足实际应用的要求.