基于人工神经网络的传感器非线性拟合方法的研究  

Study on nonlinear fitting method of transducer based on artificial neural network

作　　者：张丹[1] 于朝民[1] 付永杰[1] 

机构地区：[1]中国人民解放军92493部队

出　　处：《工业计量》2004年第5期31-33,共3页Industrial Metrology

摘　　要：人工神经网络具有以任意精度逼近任何非线性连续函数的能力。文章分别用BP网络、RBF网络对传感器的非线性作高精度拟合。仿真结果表明 ,三层BP网络和RBF网络能够满足工程实际中一维数据拟合的要求 。The artificial neural network can approximate any nonlinear function at arbitrary precision.This paper uses back propagation and RBF network to approximate the nonlinear of transducer at high precise.The approved BP algorithm speeds up learning.The simulation result shows three-1ayered BP network and RFB network can meet with one dimension curve fining in engineering practice.The network have a great generalization ability.

关 键 词：人工神经网络 非线性 传感器 拟合方法 RBF网络 BP网络 连续函数 仿真结果 数据拟合 工程实际 泛化能力 高精度 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]