基于条件生成对抗网络的信道环境缺失数据重建方法  

A Channel Environment Missing Data Reconstruction Method Based on Conditional Generative Adversarial Networks

作　　者：罗雨寒 赵灵锴 Luo Yuhan;Zhao Lingkai(Yibin University,Yibin Sichuan 644000,China)

机构地区：[1]宜宾学院,四川宜宾644000

出　　处：《现代工业经济和信息化》2024年第6期183-185,共3页Modern Industrial Economy and Informationization

摘　　要：缺失数据重建是信道环境数据预处理中的重要环节,其重建效果直接关系到信道环境监测及维护质量,但现行方法重建效果并不理想,在实际应用中数据重建误差为零占比比较小,而且数据重建速率比较低,为此提出基于条件生成对抗网络的信道环境缺失数据重建方法。采用描述统计法识别信道环境数据集中缺失数据,建立条件生成对抗网络,利用对抗网络对信道环境数据训练,提取到缺失数据特征,选择与实际情况最贴近的生成数据对缺失数据重建,实现基于条件生成对抗网络的信道环境缺失数据重建。实验证明,设计方法数据重建误差为零占比得到了有效的提升,并且重建速率也得到了有效的提升,在信道环境缺失数据重建方面具有良好的应用前景。Missing data reconstruction is an important part of channel environment data preprocessing,and its reconstruction effect directly affects the quality of channel environment monitoring and maintenance.However,the current methods for reconstruction are not ideal.In practical applications,the proportion of data reconstruction errors is relatively small,and the data reconstruction rate is relatively low.Therefore,a channel environment missing data reconstruction method based on conditional generative adversarial networks is proposed.Using descriptive statistics to identify missing data in the channel environment dataset,establishing a conditional generative adversarial network,training the channel environment data using the adversarial network,extracting missing data features,selecting the generated data that is closest to the actual situation to reconstruct the missing data,and achieving channel environment missing data reconstruction based on the conditional generative adversarial network.Experimental results have shown that the design method has effectively improved the proportion of data reconstruction errors to zero,and the reconstruction rate has also been effectively improved.It has good application prospects in the reconstruction of missing data in channel environments.

关 键 词：条件生成对抗网络 信道环境 缺失数据 数据重建 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]