基于改进经验模态分解的水电机组振动信号故障特征提取  

Fault Feature Extraction of Vibration Signals of Hydropower Units Based on Improved Empirical Mode Decomposition

作　　者：潘天航 蔡金华[1] 高元 张冰 孟宪宇[1] 耿欣[1] 冯康康[1] PAN Tian-hang;CAI Jin-hua;GAO Yuan;ZHANG Bing;MENG Xian-yu;GENG Xin;FENG Kang-kang(NR Electric Co.,Ltd.,Nanjing 211102,China)

机构地区：[1]南京南瑞继保电气有限公司,江苏南京211102

出　　处：《水电能源科学》2025年第3期182-185,162,共5页Water Resources and Power

基　　金：江苏省科技项目(BE2020688)。

摘　　要：水电机组振动信号包含噪声及多种信号耦合,采用经验模态分解方法进行信号处理会由于极值点丢失而发生模态混叠。为此,提出一种基于改进经验模态分解算法,即首先对振动信号进行微分计算来放大极值点;然而对微分信号进行经验模态分解并计算各阶信号能量特征占比,由此剔除噪声干扰等无效分量后对信号进行重构;最后计算出信号的频率特性进行故障诊断。使用该算法对国内某水电站振动异常的摆度信号进行分析,能够有效屏蔽掉干扰信号,更准确地判断出故障类型。The vibration signal of hydropower unit contains noise and a variety of signal coupling.When the empirical mode decomposition method is used to process the signal,the mode aliasing will occur due to the loss of extreme points.This paper presents an improved empirical mode decomposition algorithm for fault feature extraction of hydropower units.Firstly,the vibration signal is differentiated to amplify the extreme point.Then the differential signal is decomposed by empirical mode and the energy characteristic ratio of each order signal is calculated.Finally,the signal is reconstructed after eliminating invalid components such as noise interference.The frequency characteristic of the signal is calculated for fault diagnosis.This algorithm is used to analyze the pendulum signal of vibration anomaly of a hydropower station in China,which can effectively shield interference signals,and the fault type can be determined more accurately.

关 键 词：水电机组 微分运算 经验模态分解 故障诊断 

分 类 号：TV734.21[水利工程—水利水电工程]