摘要
C4ISR系统人机交互性能的LMBP神经网络评估模型为输入层、隐层及输出层三层前向神经网络,采用变形牛顿法,使非线性函数的平方和函数最小化。将输入提交网络,计算其网络输出和误差、敏感度、及平方误差之和。通过构造学习样本、训练网络,用Matlab工具箱仿真实现其性能评估。
The performance evaluation model of man-machine interaction of C^4ISR system based on the LMBP neural network is a forward NN with input layer, the output layer and the hidden layer. The modified Newton method is used to solve the minimal sum of squares of the nun-linier function. First, put the input data into the input layer, and then, calculate the output, the error, the sensitivity and the sum of the square error. By constructed learning stylebook and training network, and use the MATLAB toolbox to emulate to realize performance evaluation model of man-machine interaction.
出处
《兵工自动化》
2005年第4期3-4,共2页
Ordnance Industry Automation