摘要
对于一般的传感器而言 ,其输出特性容易受多种因素的干扰 ,采用传统的方法处理效果不明显或成本太高。利用神经网络的非线性映射及泛化功能 ,就能很好解决上述问题。但由于传统的 BP网络本身存在着一些缺陷。本文将通过对比不同的神经网络、不同学习算法找到一种较快收敛速度及较高精度的训练方法和神经网络。
For average sensor,the character of its output is easily interfered by various factors.The traditional ways are either inefficient or costly.With the non linearity mapping of neural network,this problem can be well handled.But the traditional BP network has demerits in itself.By comparing different neural networks and studding algorithms the following paragraph will discover a training method and neural network with high convergent speed and great accuracy.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2002年第3期298-301,共4页
Chinese Journal of Scientific Instrument
