基于深度压缩网络的电力时间序列异常快速检测  

Time series anomaly data rapid detection based on deep compression network

作　　者：卢丹 张琳娟 李翼铭 许长清 孙合法 翁理国[2] LU Dan;ZHANG Lin-juan;LI Yi-ming;XU Chang-qing;SUN He-fa;WENG Li-guo(State Grid Henan Economics Research Institute,Zhengzhou 450052,China;School of Automation,Nanjing University of Information Science and Technology,Nanjing 210044,China)

机构地区：[1]国网河南省电力公司经济技术研究院,郑州450052 [2]南京信息工程大学自动化学院,南京210044

出　　处：《信息技术》2025年第4期97-104,109,共9页Information Technology

摘　　要：电力时间序列异常数据检测是电力领域的研究热点,也是数据分析的前提任务。机器学习方法未能很好提取数据关联性及发展趋势,深度学习方法由于参数量多导致计算时间过长、部署困难。因此,文中提出一种压缩网络模型,利用深层信息提取能力预测数据发展趋势,对数据进行异常检测;在训练模型时同步进行压缩与剪枝等剔除冗余参数,减少模型大小,提升运行速度,在精度与速度上同步提升。最后,将模型输出结果输入轻量梯度提升树中进一步提升精度。实验结果表明,文中方法可以在有效识别异常数据的前提下大幅度减少计算时间。The detection of abnormal data in power time series is a research hotspot in the field of power and a prerequisite task for data analysis.Machine learning methods have not been able to extract data correlations and development trends effectively,while deep learning methods suffer from long computation time and deployment challenges due to a large number of parameters.Therefore,this paper proposes a compressed network model that utilizes the deep information extraction capability to predict data development trends and detect anomalies.During model training,redundant parameters are simultaneously removed through compression and pruning techniques to reduce model size and improve runtime speed,while maintaining accuracy and speed simultaneously.Finally,the model output is further enhanced by inputting it into a lightweight gradient boosting tree to improve accuracy.Experiment results demonstrate that the proposed method can significantly reduce computation time while effectively identifying abnormal data.

关 键 词：深度压缩网络 时间序列 异常检测 剪枝 梯度提升树 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]