整数微分进化策略及其在微波成像中的应用  

Integer Coded Differential Evolution Strategy and Application to Microwave Imaging

作　　者：廖成[1] 卫涛[1] 陈伟[1] 

机构地区：[1]西南交通大学电磁所,四川成都610031

出　　处：《西南交通大学学报》2007年第6期647-652,共6页Journal of Southwest Jiaotong University

基　　金：国家自然科学基金资助项目(10476021)

摘　　要：提出了微分进化策略的整数编码方法.按照给定的精度使优化问题的常规实数编码的解空间离散化,并对离散化的结果进行整数编码,建立对应的整数空间.在整数空间中求解,在实数空间中计算解的适应度.对有限精度以及计算存在离散化过程的工程优化问题,用整数编码可有效提高搜索速度,并自动滤除不稳定解.对于离散、高维、多值类型的微波成像示例,与实数编码比较,用微分进化策略的整数编码优化算法搜索速度更高,成像结果更优.Integer coding for a differential evolution strategy was proposed. For an optimization problem, the conventional solution space coded with real numbers is discretized into an integer coded space according to a given precision. Solutions were searched in the integer coded space while the fitness is evaluated in the real space. For engineering problems, the integer coding is efficient, and automatically filters unsteady solutions. As an example, the integer coded differential evolution strategy was applied to solve a microwave imaging problem, which was of discretization, high-dimension and multi-solution. The integer coded differential evolution strategy shows a higher searching speed and obtains better optimization results compared with the real coding.

关 键 词：微分进化策略 微波成像 整数编码 实数编码 全局优化 

分 类 号：TN822.8[电子电信—信息与通信工程]