摘要
针对雷达阵地毁伤评估问题,提出了一种基于GA-动态BP神经网络的评估方法。该方法首先通过遗传算法(GA)寻求神经网络最优权重值,充分发挥了动态神经网络的泛化和非线性映射能力以及GA的全局寻优能力,提高了结果的精确度;再引入牛顿迭代法优化网络训练算法,克服了神经网络在训练过程中易陷入局部极小值及网络学习后期收敛速度慢的缺点。以某一时刻防空作战为想定,仿真实现了雷达阵地的毁伤评估,与现有算法相比,该算法在收敛速度、可靠性和准确性上都有明显提高。
Aimed at the problem of radar position damage assessment, a GA- dynamic BP neural network method is proposed. Firstly, GA is used to find the optimal weights of neural network, taking full advantage of generalization and nonlinear mapping ability of dynamic neural network and global optimization ability of GA, which improves the accuracy of results. In addition, Newton iterative method is introduced to optimize traditional network training algorithm, which overcomes the disadvantage of easily falling into local minimum and the low convergence rate in the course of training. Taking a moment in air defense combat, achieve a simulation of radar position damage assessment. Compared with existed algorithm, this algorithm has a significant improvement in convergence speed, reliability and accuracy.
出处
《战术导弹技术》
北大核心
2016年第6期51-57,共7页
Tactical Missile Technology
基金
武器装备军内科研项目(KJ2013122)
关键词
毁伤评估
雷达阵地
遗传算法
神经网络
牛顿迭代法
damage assessment
radar position
genetic algorithm
neural network
newton iterative method
