浙江电力

2021, v.40;No.301(05) 36-44

[打印本页] [关闭]
本期目录(Current Issue) | 过刊浏览(Archive) | 高级检索(Advanced Search)

基于边缘侧自编码器压缩的换流站设备故障预测
Fault Prediction of Converter Station Equipment Based on Edge Side Autoencoder Compression

胡宪,张彩友,冯毅萍,顾天雄,潘戈,富银芳
HU Xian,ZHANG Caiyou,FENG Yiping,GU Tianxiong,PAN Ge,FU Yinfang

摘要(Abstract):

随着电力需求的日益增长,保证电力设备维持在高可用的状态变得非常重要。故障预警通过预测设备的潜在故障,协助运维人员提前定位系统不稳定因素并加以干预,成为设备主动运维的重要方法。结合换流站设备呈区域分布的特点,提出了一种基于边缘侧自编码器压缩的分布式LSTM(长短期记忆网络)预测模型。该模型将生产区域抽象为边缘节点,通过自编码器提取表征各边缘节点设备状态的编码信息,并利用工业网络进行信息共享。通过特征工程的方法改进了传统LSTM模型的输入,在提升模型预测精度的同时降低了模型复杂度。某换流站的应用案例验证了方法的有效性和可行性。
With the growing demand for electric power, it is of great consequence to maintain high availability of power equipment. Fault warning, by prediction of potential equipment faults, can help maintenance personnel in advance locate and intervene in unstable factors and now becomes an active maintenance method. In view of the regional distribution of converter stations, the paper proposes a distributed LSTM(long short-term memory) prediction model based on edge side autoencoder compression which abstracts the production area as an edge node, utilizes the autoencoder to extract the coded information that characterizes the equipment status of edge device and shares the information through industrial networks. By means of featuring engineering, the input of the traditional LTSM model is improved, the model prediction precision raised and model complexity reduced. The effectiveness and feasibility of the proposed method are verified by a case study of a converter station.

关键词(KeyWords): 边缘侧;分布式计算;故障预测;自编码器
edge side;distributed computing;fault prediction;autoencoder

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目(5211MR18004N)

作者(Author): 胡宪,张彩友,冯毅萍,顾天雄,潘戈,富银芳
HU Xian,ZHANG Caiyou,FENG Yiping,GU Tianxiong,PAN Ge,FU Yinfang

DOI: 10.19585/j.zjdl.202105006

参考文献(References):

扩展功能
本文信息
服务与反馈
本文关键词相关文章
本文作者相关文章
中国知网
分享