基于改进神经网络的光纤通信故障数据的自动识别  

Automatic Recognition based on Improved Neural Networks Optical Data Communication Failure

作　　者：邓荣[1] 唐林[1] 

机构地区：[1]重庆工程职业技术学院信息工程学院,重庆402260

出　　处：《激光杂志》2016年第6期142-146,共5页Laser Journal

基　　金：重庆市教委科学技术研究项目(项目编号:152072)

摘　　要：针对传统神经网络构建的光纤通信网络故障数据识别器,在训练进程中易出现局部突出极小值,导致故障数据识别不准确的问题。对光纤的内部结构和通信故障数据的自动识别问题进行研究,提出一种基于免疫识别的神经网络光纤通信网络故障自动识别模型,分析传统的神经网络在识别进程中的弊端,将改进的神经网络结构同免疫识别理论相结合重新构建神经网络自动识别器,并应用于光纤通信网络中,用给出的相应训练方法进行训练,把光纤通信网络的故障数据存储在识别器中,获取故障模式特征,依据识别器的激活状态自动识别光纤通信网络的故障点。实验结果表明,本文提出的基于免疫识别的神经网络光纤通信网络故障检测方法,可准确识别出光纤通信网络故障数据,其识别率明显高于传统的识别方法。Optical communication failure automatic data recognition problems. For optical communication networks prone prominent local minima lead to failure to identify inaccurate defects based on traditional neural network training process. Immune recognition presents a neural network optical communication network failure automatic recognition model for the shortcomings of existing neural networks generated in the identification process for analysis,the improved neural network structure combined with immune recognition of the theory of neural networks for building automatic recognizer,based on the training data stored in the fault in the optical communication network identifier,the identifier is applied to obtain the characteristic failure mode optical communications network,according to the active state recognizer automatically identify the point of failure in the optical communication network,and presents appropriate training methods. Experimental results show that the proposed method can accurately identify the optical communication network failure data,and has a high recognition rate.

关 键 词：改进神经网络 光纤通信 故障数据检测 自动识别 

分 类 号：TP274.4[自动化与计算机技术—检测技术与自动化装置]