动态聚类适应值共享遗传算法在散射计风矢量反演中的应用  

Application of dynamic-clustering-based fitness-sharing genetic algorithm to wind vector retrieval for scatterometer

作　　者：解学通[1] 郁文贤[1] 郭丽青[2] 方裕[3] 

机构地区：[1]国防科学技术大学电子科学与工程学院 [2]中山大学教学实验中心,广州510275 [3]北京大学遥感与地理信息系统研究所,北京100871

出　　处：《计算机应用》2008年第1期97-100,共4页journal of Computer Applications

基　　金：国家863计划项目(2002AA134100);教育部985工程项目(105203200400006)

摘　　要：风矢量反演是散射计数据处理的核心内容,传统风矢量反演算法的设计过多依赖于目标函数的具体分布形态。以SeaWinds散射计为例,根据风矢量反演的多解问题和模糊解特性,设计了一种基于动态小生境遗传算法的风矢量反演算法。利用部分L2A和相应L2B数据对该算法进行了验证。结果表明该算法在无需任何目标函数先验知识的条件下能够取得较好的反演结果。Wind vector retrieval is the key to scatterometer data processing. The design of traditional wind vector retrieval algorithm excessively depends on the detailed distribution shape of the objective function. Taking SeaWinds as an example, according to the multiple solution problem of wind vector retrieval and its ambiguity solution property, a wind vector retrieval algorithm was designed, which was based on the dynamic clustering niche genetic algorithm. By using some Level 2A data and the corresponding Level 2B data, the algorithm was validated. The experimental results indicate that the algorithm achieves good wind field results without using any transcendental knowledge of the objective function.

关 键 词：散射计 风矢量反演 聚类 适应值共享 遗传算法 

分 类 号：TP722.6[自动化与计算机技术—检测技术与自动化装置]