基于概率神经网络的光纤智能结构承载定位  

Load Localization of Optical Fiber Smart Structures Based on Probabilistic Neural Networks

作　　者：李鹏[1] 

机构地区：[1]华东交通大学机电工程学院,江西南昌330013

出　　处：《广西师范大学学报（自然科学版）》2011年第2期223-226,共4页Journal of Guangxi Normal University:Natural Science Edition

基　　金：国家自然科学基金资助项目(61063037);华东交通大学校立课题(09102005)

摘　　要：本文提出一种用于识别光纤智能结构承载位置的方法,研究构建光纤传感网络,设计相应的监测系统,获得用于表征结构承载信息的特性向量。同时,运用神经网络方法构造了四层结构的概率神经网络,并以特征向量作为样本,对结构的承载定位进行研究,重点分析训练样本数和平滑系数对定位效果的影响。实验结果表明:采用训练样本数为30,平滑系数为0.25的概率神经网络对承载位置的定位正确率可达93.2%,研究最终给出用于承载定位的概率神经网络模型。In this paper,the load location of optical fiber smart structures is identified based on probabilistic neural networks. The optical fiber network and the monitoring circuit were designed to acquire structural load information as feature vectors. The probabilistic neural network structure is brought forward by employing the method of neural network. The feature vectors as the sample is presented to the probabilistic neural network. The effect of smoothing parameter and number of training samples on the localization performance of the probabilistic neural network is studied. The result shows that good selection of the number of training samples （30） and smoothing parameter （0. 25） boost the load localization accuracy （93.2%）. The localization model of probabilistic neural was presented finally.

关 键 词：光纤智能结构 承载定位 概率神经网络 平滑系数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]