基于遗传算法的工具误差分离与弹道折合  

Instrumentation Errors Separation and Trajectory Conversion Based on Genetic Algorithm-Principal Component

作　　者：夏青[1] 杨华波[1] 张士峰[1] 蔡洪[1] 

机构地区：[1]国防科技大学航天与材料工程学院,湖南长沙410073

出　　处：《系统仿真学报》2007年第18期4130-4133,共4页Journal of System Simulation

摘　　要：制导工具误差分离与折合是导弹精度评定中的重要问题,针对传统主成分方法中主要成分选择上的困难,提出了利用遗传算法确定最佳主成分子集的方法,遗传算法适应度函数综合考虑了模型拟合与预测性能,能够有效避免过拟合问题,使得分离结果的预测能力显著增强。仿真计算表明,与最小二乘和主成分方法相比,遗传算法获得的误差系数估计与真值更加接近,折合得到的全程弹道遥外差更加准确。The separation and conversion of guidance instrumentation errors plays a vital role in missile accuracy assessment. For the problem of selecting principal components in traditional principal component analysis, the genetic algorithm was used to select an optimal principal component set. The fitness function of genetic algorithm synthetically considered model fitting and forecast capability, which could effectively avoid over-fitting, and enhance forecast capability of separation result. The results show that the new method has a higher separating and converting precision than least square method and principal component method.

关 键 词：工具误差分离 主成分估计 遗传算法 弹道折合 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]