基于广义回归神经网络的传感器非线性误差校正  

Nonlinear error correction of sensor based on generalized regression neural network

作　　者：段松杰[1] 张晓光[1] 张闯志[1] 

机构地区：[1]中国矿业大学机电工程学院,江苏徐州221008

出　　处：《传感器与微系统》2008年第12期14-16,共3页Transducer and Microsystem Technologies

基　　金：中国博士后科学基金资助项目(20060390277);江苏省博士后科研基金资助项目(0502010B);江苏省"六大人才高峰"基金资助项目(06-E-052);江苏省工业高技术研究计划资助项目(BG2007013)

摘　　要：介绍了径向基函数网络的函数逼近原理和方法,提出了一种基于广义回归神经网络(GRNN)的传感器非线性误差校正方法。通过Matlab的Network Toolbox(神经网络工具箱),GRNN训练程序实现了输出特性曲线逼近。仿真分析表明:GRNN能够很好地满足传感器非线性拟合的要求,网络结构简单,收敛速度快。The RBF network function approximation theory and method are introduced, and the method of nonlinear error correction of sensor is presented based on generalized regression neural network（GRNN）. By the neural Network Toolbox in Matlab a generalized regression neural network training program fulfills approximation towards the output characteristic curve. The simulation results show that GRNN can meet the requirements of nonlinear fitting of sensors, and the structure of network is simple, with fast convergence.

关 键 词：广义回归神经网络 传感器 非线性误差 径向基函数 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]