摘要
空袭目标优选,应用改进的BP神经网络算法。即在空袭目标优选评估指标体系基础上,建立BP神经网络,通过定义学习代价函数、确定函数输出信号、修正函数信号。并通过网络初始化等改进BP算法,以激活函数敏感性,加速网络收敛。算例验证了该算法合理性,稳定性良好。
The optimal selection of air raid objects adopts modified BP NN algorithm. That is, based on the optimal selection of air raid objects evaluation index system, modified BP NN algorithm was built, while the algorithm ensure function output signal and modify function signal through define study cost function. Modify the BP algorithm through network initialization to active function sensitivity and accelerate network convergence. The results proved the rationality and stability of models.
出处
《兵工自动化》
2007年第6期5-7,共3页
Ordnance Industry Automation
基金
高等学校骨干教师资助计划(GG-1105-90039-1004)
关键词
空袭目标
目标优选
BP神经网络
改进BP算法
网络初始化
函数敏感性
Air raid objects
Optimal selection of objects
BP NN
Modified BP NN algorithm
Network initialization
Function sensitivity
