摘要
人工神经网络具有以任意精度逼近任何非线性函数的能力.本文分别用BP网络、RBF网络对压力传感器的静态特性作高精度拟合,改进的BP算法加速了网络的收敛.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,网络具有良好的泛化能力.
The artificial neural networks can approximate any nonlinear function at arbitrary precision.This paper uses backpropagation and RBF networks to approximate the static characteristics of pressure transducer at high precise.The approved BP algorithm speeds up learning.The simulation result shows three\|layered BP network and RFB network can meet with one dimension curve fitting in engineering practice.The networks have a great generalization ability.
出处
《装备指挥技术学院学报》
2002年第3期74-76,共3页
Journal of the Academy of Equipment Command & Technology
