基于改进VMD和GRU的水轮发电机组振动故障预警  

Vibration Fault Warning of Hydroelectric Generating Sets Based on Improved VMD and GRU

作　　者：皮有春 谭鋆[1] 郭钰静[1] 黄正海 肖燕凤[1] 陈言 PI You-chun;TAN Yun;GUO Yu-jing;HUANG Zheng-hai;XIAO Yan-feng;CHEN Yan(China Yangtze River Electric Power Co.,Ltd,Yichang 443000,Hubei Province,China;Shanghai Changgeng Information Technology Co.,Ltd,Pudong,Shanghai 201209,China)

机构地区：[1]中国长江电力股份有限公司,湖北宜昌443000 [2]上海长庚信息技术股份有限公司,上海201209

出　　处：《中国农村水利水电》2024年第3期244-249,共6页China Rural Water and Hydropower

基　　金：水轮发电机组状态趋势预警高级应用研发(1520020007)。

摘　　要：针对水轮发电机组受水、机、电等因素相互耦合,早期故障特征被电磁和噪声所淹没难以提取的问题,设计了一种基于改进VMD和GRU的水轮发电机组振动故障预警方法。首先,采用BES算法对变分模态VMD的参数进行寻优,得到最佳的分解层数、惩罚因子和模态个数,然后采用最优的VMD算法对水轮发电机组早期的振动特征进行提取,最后将早期的振动特征输入GRU神经网络预测算法进行训练、验证和测试。仿真结果和工程实例表明,该方法可以有效快速准确提取水轮发电机组的早期微弱振动特征,实现水轮发电机组的早期故障预警,具有较高的工程应用价值。In view of the problem of water,machine,electricity and other factors,and the early fault characteristics submerged by electromagnetic and noise,a vibration fault warning method of hydroelectric generating sets based on improved VMD and GRU is designed in this paper.First,the BES algorithm is used to optimize the parameters of variational mode VMD to obtain the optimal number of decomposition layers,penalty factor and mode number.Then,the optimal VMD algorithm is adopted to extract the early vibration fault features of hydroelectric generating sets,and finally,the extracted early vibration features are input into the GRU neural network prediction algorithm for training,validation and testing.Simulation results and engineering examples show that this method can extract the early weak vibration characteristics of hydroelectric generating sets effectively and quickly and accurately,and achieve the early fault warning of hydroelectric generating sets,which has high engineering application value.

关 键 词：水轮发电机组 BES VMD GRU 振动故障 

分 类 号：TV7[水利工程—水利水电工程] TK73[交通运输工程—轮机工程]