基于EMD与小波包分析的燃气轮机故障诊断方法研究
Research on Gas Turbine Fault Diagnosis Method Based on EMD and Wavelet Packet Analysis
陈维钧1
作者信息
- 1. 深圳大唐宝昌燃气发电有限公司,深圳 518110
- 折叠
摘要
燃气轮机转子通常在极高转速和高温环境下运行,在运行过程中产生的振动信号是非平稳的,使得故障特征信号容易被噪声掩盖,故障的细微特征难以被捕捉到,导致燃气轮机故障诊断精度降低.为此,引入经验模态分解(Empirical Mode Decomposition,EMD)与小波包分析方法,提出新的燃气轮机故障诊断方法.采用EMD方法将燃气轮机转子信号分解为高低频本征模态函数(Intrinsic Mode Function,IMF)分量,获取转子信号IMF能量特征,引入小波包分析提取不同尺度上的小波能谱熵;将转子信号特征输入至中,构建最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)决策函数,用于诊断燃气轮机故障.仿真结果表明,经过所提方法处理后,振动信号中的燃气轮机故障信号位置变得清晰可辨,能够有效识别出实际发生的故障.
Abstract
Gas turbine rotors usually operate at extremely high speed and high temperature,and the vibration signals generated during operation are non-stationary,which makes the fault characteristic signals easy to be covered by noise,and the fine fault characteristics difficult to be captured,resulting in the reduction of gas turbine fault diagnosis accuracy.Therefore,a new fault diagnosis method of gas turbine is proposed by introducing Empirical Mode Decomposition(EMD)and wavelet packet analysis.EMD method is used to decompose gas turbine rotor signals into high and low frequency Intrinsic Mode Function(IMF)components,obtain the IMF energy characteristics of rotor signals,and introduce wavelet packet analysis to extract the wavelet spectrum entropy at different scales.The rotor signal characteristics were input into,and the Least Squares Support Vector Machine(LSSVM)decision function was constructed to diagnose gas turbine faults.The simulation results show that after the proposed method is processed,the location of the gas turbine fault signal in the vibration signal can be clearly identified,and the actual fault can be effectively identified.
关键词
燃气轮机/故障诊断/经验模态分解(EMD)/小波包分析Key words
gas turbine/fault diagnosis/Empirical Mode Decomposition(EMD)/wavelet packet analysis引用本文复制引用
出版年
2024
NETL
NSTL
万方数据