摘要
研究了采用BP神经网络实现传感器逆向建摸,用三种神经网络方法(LM算法)计算比较了使用两种不同初始化规则和不同网络结构时对网络性能和计算精度的影响。计算机仿真实验表明:使用NW初始化规则并改进网络结构后,网络的收敛速度更快,精度更高。
The inverse modeling of sensor based on BP neural network is adopted and three methods(LM algorithm) are introduced.Two different initial rules and different network architectures are compared in the aspect of network performance and precision.The results of computer simulation illustrate that the method which uses Nguyen-Widrow initial rule and improved network architecture has a high converging speed and good precision.
出处
《传感器技术》
CSCD
北大核心
2005年第2期11-12,共2页
Journal of Transducer Technology
基金
国家自然科学基金资助项目(30170240)
西安市科技攻关项目基金资助项目(GG04038)
关键词
BP神经网络
LM算法
NW初始化规则
传感器
非线性误差
BP(back propagation) neural network
LM(Levenberg-Marquardt) algorithm
NW(Nguyen-Widrow) initial rule
sensor
nonlinear errors