摘要
动能杆侵彻靶板的毁伤效果评估是一个复杂的问题。由于动能杆的数量多,侵彻碰撞情况多达上千种,计算费力又耗时。针对这些问题,提出了一种评价毁伤效果的数学模型,建立了基于反向传播(back propagation,BP)神经网络的预测模型,给出了隐含层节点数的计算方法,并在网络节点作用函数和随误差的变化动态更改参数方面对BP神经网络进行了改进。最后对仿真结果进行了分析,结果表明,75%的神经网络预测值分布在±10%误差范围之内,86%的神经网络预测值分布在±20%误差范围之内,验证了预测模型的可靠性和有效性。
The assessment of damage effect for kinetic energy rod （KE-rod） penetration on targets is a com plex problem, and the computation is laborious and time-consuming because of a large number of KE rods and thousands of penetrations. Aiming at these problems, a mathematical model is put forward to evaluate the dam- age effect, the forecast model based on back propagation （BP） neural network is established, the calculative method of hidden layer nodes is given and the BP neural network is improved on the network function node and the dynamic changes of parameters with the error. The simulation results show that 75% of the predicted value based on neural network is distributed with ±10% error and 86%o of the predicted value based on neural net work is distributed with ＋20% error, which validates that the forecast model is reliability and validity.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2013年第9期1898-1902,共5页
Systems Engineering and Electronics
基金
航空科学基金(20120196006)资助课题
关键词
毁伤指数
动能杆
反向传播神经网络
侵彻
damage index
kinetic energy rod
back propagation （BP） neural network
penetration
