Vibration Signal Feature Extraction of Hydropower Unit Based on Wavelet Packet Decomposition and CEEMDAN Energy Entropy
Aiming at the problems of non-stationary nonlinear vibration signal of hydropower unit with excessive noise signal,this paper proposes a feature extraction method based on the combination of adaptive noise complete empirical mode decomposition(CEEMDAN)and energy entropy.Firstly,the collected vibration signal was de-noised by wavelet packet decomposition,and the de-noised signal was decomposed by CEEMDAN.Then the effective intrinsic mode func-tion(IMF)was screened by correlation coefficient method,and the energy entropy of these IMF components was calcu-lated to construct the eigenvector set,which was finally input into the Marine Predator optimization support vector ma-chine(MPA-SVM)for pattern recognition.By using the simulated signal and the measured signal,the proposed method was compared with other methods.The experimental results show that the feature extraction method based on wavelet packet decomposition and CEEMDAN energy entropy can accurately extract features and effectively distinguish different states of units,which provides application value for engineering field.
万方数据
维普
