摘要
以往针对通信网络故障分类的算法没有考虑告警和故障数据中的潜在特征,导致故障分类准确率低,因此提出一种基于数据挖掘的通信网络故障分类算法。首先,根据对数据背景和数据特点的理解,使用特征构造挖掘数据中潜在的特征,将挖掘到的特征加入原数据中。然后,使用LightGBM算法的特征重要性评估函数对新数据集中的所有特征进行重要性评估,根据重要性值删除不重要特征。最后,使用集成学习模型对特征筛选后的数据集进行故障分类研究。实验结果表明,基于数据挖掘的通信网络故障分类算法的准确率有更好的效果。
The communication network fault classification algorithms previously used did not consider potential features in alarm and fault data, resulting in low classification accuracy. This study proposes a data mining-based communication network fault classification algorithm to address this issue. First, the feature structure is applied to the mining of the potential features in the data based on the understanding of the data background and data characteristics, and then the mined features are added to the original data. Furthermore, the LightGBM algorithm’s feature importance evaluation function is used to evaluate the importance of all features in the new dataset and to delete unimportant features based on the importance value. Finally, ensemble learning algorithms are used to perform fault classification studies on featurescreened datasets. The experimental results show that the accuracy of the communication network fault classification algorithm based on data mining has a better effect.
作者
朱圳
刘立芳
齐小刚
ZHU Zhen;LIU Lifang;QI Xiaogang(School of Computer Science and Technology,Xidian University,Xi'an 710071,China;School of Mathematics and Statistics,Xidian University,Xi'an 710071,China)
出处
《智能系统学报》
CSCD
北大核心
2022年第6期1228-1234,共7页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(61877067)。
关键词
数据挖掘
通信网络
故障分类
集成学习
特征工程
告警数据
网络故障
网络告警
data mining
communication network
fault classification
integrated learning
characteristic engineering
alarm data
network failure
network alarm
