遗传神经网络在求解初等函数高阶导数值中的应用

传统的求解函数高阶导数值的方法就是先求出高阶导数的函数表达式,然后将自变量值代入,就得到了此点的高阶导数值。高阶导数的函数表达式的推导比较的烦琐,尤其对于复合函数来说。利用改进的遗传算法和神经网络各自的优点,提出求解函数高阶导数值的GA-Network法。算法采用多目标优化的思想,使用"动态自适应策略"和"罚函数法"。利用神经网络来构造函数泰勒展式的网络结构,用遗传算法对网络进行学习,最后得到网络的输出结果即高阶导数值。通过对初等函数的仿真实验,可以看出此方法有比较高的精度,它也为函数导数值的求解提供了一种方法。

Genetic Algorithm and Neural Network to Solve the Simple Function's Derivative of Higher Order

The traditional method of solving higher derivative is to derive higher-order derivative expression,and then the variable will be replaced by the value,the high-order derivative has been derived on this point. The process for deducing the function expression of higher-order derivative is complex,especially for some composite function. In this paper,the strong points of advanced genetic algorithm (GA) and neural network were used,the GA-Network to solve the function's derivative of higher order was introduce...