基于数据降维与神经网络的计算机通信网络故障识别方法  

Fault Identification Method for Computer Communication Networks Based on Data Dimensionality Reduction and Neural Networks

作　　者：陈凯锋 CHEN Kaifeng(China Mobile Tietong Sichuan Branch,Chengdu Sichuan 630014,China)

机构地区：[1]中移铁通四川分公司,四川成都630014

出　　处：《信息与电脑》2024年第24期139-141,共3页Information & Computer

摘　　要：针对计算机通信网络中常见的故障识别问题,文章提出了一种基于数据降维技术和改进卷积神经网络的方法。该方法首先通过主成分分析对高维故障数据进行降维处理,然后构建并训练卷积神经网络模型,以自动识别网络故障。实验结果表明,改进卷积神经网络方法在识别准确率和计算效率方面均优于传统小波分析方法,尤其在大规模数据集上表现突出。结果指出,基于改进卷积神经网络的方法在提高识别准确率、减少计算时间和降低模型复杂度方面具有明显优势,在现代网络运维中具有广阔的应用前景。The paper proposes a method based on data dimensionality reduction and improved convolutional neural networks for fault recognition in computer communication networks.The method first applies principal component analysis to reduce the dimensionality of high-dimensional fault data,then constructs and trains a convolutional neural network to identify network faults.Experimental results show that the improved convolutional neural network outperforms traditional wavelet analysis methods in both accuracy and efficiency,especially on large-scale datasets.The results indicate that this method improves recognition accuracy,reduces computation time,and simplifies the model,making it highly applicable to modern network operation and maintenance.

关 键 词：数据降维 神经网络 计算机通信网络 故障识别 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]