基于遗传算法的曲面拟合参数辨识  

Parameter Identification of Surface Fitting Based on Genetic Algorithm

作　　者：谷川[1,2] 潘国荣[1,3] 施贵刚[1] 陈兴权[1] 

机构地区：[1]同济大学测量与国土信息工程系,上海市四平路1239号200092 [2]上海市政工程设计研究总院,上海市中山北二路901号200092 [3]现代工程测量国家测绘局重点实验室,上海市四平路1239号200030

出　　处：《武汉大学学报（信息科学版）》2009年第8期983-986,991,共5页Geomatics and Information Science of Wuhan University

基　　金：国家科技支撑计划资助项目(2006BAJ03A07)

摘　　要：从参数辨识的角度来看,曲面拟合需要求解出坐标平移、旋转以及标准曲面方程参数。利用遗传算法在该领域的优势,对经典简单遗传算法的缺陷进行了一系列改进。用MATLAB语言实现了改进算法的程序包。在雷达天线椭圆抛物面表面检测的工程实例以及一个模拟复杂曲面算例中,分别采用文章中的改进遗传算法和简单遗传算法进行了多次运算,对结果进行了比较。应用和比较结果表明,遗传算法能够较好地应用于空间曲面拟合,且本改进算法更具优势。Viewing from the aspect of parameter identification, the parameters of coordinate translation, rotation and the standard curved surface equation are to be estimated. Genetic algorithms have particular advantages in this field. A series of modifications are made aiming at several defects of the classic simple genetic algorithms, and an improved algorithm is realized using MATLAB. Both algorithms are used many times in an engineering examples and a simulated complex surface example, and the calculation results are compared. Applications and comparisous show that the genetic algorithm can be well used in the field of spatial surface fitting, and the improved algorithm is better.

关 键 词：曲面拟合 参数辨识 遗传算法 坐标转换 椭圆抛物面 模拟复杂曲面 

分 类 号：P34[天文地球—水文科学] P258[天文地球—地球物理学]