融合遗传算法与BP神经网络的气象威胁度建模与评估  

Weather Threat Modeling and Assessment Based on Genetic Algorithm and BP Neural Networks

作　　者：王俊[1] 周树道[1] 叶松[1] 程龙[1] 罗炜[1] 

机构地区：[1]解放军理工大学气象学院,南京211101

出　　处：《电光与控制》2012年第3期74-77,共4页Electronics Optics & Control

基　　金：国家自然科学基金(40976062)

摘　　要：针对无人飞行器航迹规划中的气象威胁要素模糊性强、复杂多变等特点,提出一种融合遗传算法与BP神经网络的气象威胁度评估方法。利用遗传算法的全局寻优能力优化BP神经网络的初始参数及结构,结合GA-BP神经网络对气象威胁度进行建模与评估。通过Matlab仿真验证,结果表明该方法能够准确评估气象威胁度,与BP神经网络相比,具有更快的收敛速度、更好的全局收敛性,提高了评估效率与准确度。Considering the obscure, complicated and changeable weather threat in route planning of Unmanned Aerial Vehicles（UAV） , the authors proposed an algorithm of weather threat level assessment by combining Genetic Algorithm （GA） with BP Neural Networks. The initial parameters and their structures of BP neural network were improved by the global optimization of Genetic Algorithm. The weather threat was modeled and assessed by GA-BP neural networks. Matlab simulation results showed that the method can assess the degree of weather threat accurately and have a faster convergence rate and better global convergence. The efficiency and accuracy of assessment were also improved.

关 键 词：无人飞行器 航迹规划 气象威胁 遗传算法 BP神经网络 评估 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] TP391.9[自动化与计算机技术—计算机应用技术]