本文已被：浏览 1036次 下载 0次
中文摘要: 本文采用BP 多层前馈神经网络及其改进算法对传感器特性进行补偿. 提出附加动量法、自适应参数变化法
为主要内容的BP 神经网络改进算法,有效地改善了BP 网络传统算法收敛慢、容易收敛到局部最小点的缺陷,并编
制了训练程序. 结果表明,经BP 改进算法处理后,传感器性能大幅度改善,改进的BP 算法拟合精度更高,计算时间
更短.
Abstract:BP neut ral network and it s improved algorithms are applied to compensate sensor′s performance. BP
neut ral network′s improved algorithms of adding momenturn and self2adapting parameter are presented. The de2
fect s of BP , for example , converging slowly , being easy to converge to minimum of one part are improved effi2
ciently. Training programs are done by using MATLAB language. Result s show that the performance of sensor
is improved highly.
keywords: BP algorithm data fusion sensor
文章编号：cg020301 中图分类号： 文献标志码：
基金项目:
ZHA N G Yonghuai L IU J unhua
Xi′an Jiaotong University , Xi′an 710049 , P. R. China
引用文本：