摘要
针对目前标准BP神经网络的缺点,提出基于高阶导数的多记忆BP算法,将能量函数的n阶导数与最速下降方向相结合,构造出一个新的最速下降方向,从而提高了神经网络的学习速度.证明了该算法相对于传统梯度算法的快速性,给出了该算法的实现方法,并进行了算例仿真.为了证明其实效性,设计了汽车半主动悬架神经网络控制器.结果证明,该算法便捷、实用、有效.
Regarding drawbacks of the standard BP algorithm algorithm is proposed. It combines the n-th order of energy , a high-order derivative based multiple memory BP function with the direction of the fastest decline to construct a new direction of the fastest decline, and improve the learning speed of the neural network. The new algorithm is compared with the traditional gradient algorithm to show its high computation speed. Implementation of the new algorithm is given. Finally a neural network controller is designed to optimize the performance of the automobile suspension. The result shows that the new algorithm is convenient and effective.
出处
《应用科学学报》
CAS
CSCD
北大核心
2008年第1期85-88,共4页
Journal of Applied Sciences
基金
广东省科技计划资助项目(No.2005B10201014)
关键词
神经网络
BP算法
高阶导数
悬架
neural network
BP algorithm
high order derivative
suspension
