论文部分内容阅读
本文利用多层前向神经网络的函数逼近能力,推导出一阶微分方程组边值问题的神经网络解法,为一阶微分方程组的近似解析解的求取提供了便于工程应用的新方法.
In this paper, by using the function approximation ability of multi-layer feedforward neural network, the neural network method for the boundary value problem of first-order differential equations is deduced, which provides a new method for engineering approximate solutions of first-order differential equations.