基于小波包滤波的汽轮机转子振动故障的Kolmogorov熵诊断

KOLMOGOROV ENTROPY DIAGNOSIS FOR VIBRATION FAULTS OF TURBINE ROTOR BASED ON WAVELET PACKET FILTERING

根据Bently实验台所采集的碰摩、松动、不对中、不平衡四种典型汽轮机转子振动故障信号,采用小波包和Kolmogorov熵相结合的方法对其进行故障诊断。即先利用小波包方法对原始信号进行滤波,提取有用的信号频段,再对滤波后的信号求解Kolmogorov熵。诊断结果表明:小波包分析的方法有着很好的滤波和提取非平稳信号的能力;小波包滤波后的汽轮机转子的振动时间序列在不同故障状态下的Kolmogorov熵明显不同,Kolmogorov熵在进行汽轮机转子故障类型诊断时有较好的区分度。

Taking into consideration the four typical fault signals of turbine rotor vibration, including rubbing, loosening, misalignment and mass unbalance, collected from the Bently experiment set, the method combining wavelet packet and Kolmogorov entropy is introduced to carry out the classification and diagnosis. The wavelet packet is used for filtering and extracting useful signal frequency segments from the original signal, and then the corresponding Kolmogorov entropy of the signal is adopted as the basis of diagnosis. The results indicate: wavelet packet is of good ability of filtering and nonstationary signal extracting. Kolmogorov entropy of vibration time series can distinguish different fault conditions, and is of good discriminability in turbine rotor faults diagnosis.