Detection And Analysis Of Subsynchronous Oscillation Based On DBSCAN-kmeans Clustering And Fourier-based Synchrosqueezing Transform

With the continuous development and expansion of the modern power system,wide application of renewable energy and power electronic equipment,the problems of subsynchronous oscillation(SSO)in power systems have changed constantly,which threatens the stability and security of the system.In the study of subsynchronous oscillation,the detection and analysis of the power system subsynchronous signal is an important part,which analyzes the oscillation phenomenon that has occurred,and provides the necessary reference for the follow-up research.In this paper,the detection and analysis method of the subsynchronous oscillation signal is studied and applied.Firstly,several classical methods of time-frequency analysis are introduced,such as Short Time Fourier Transform(STFT)based on Fast Fourier Transform,Wavelet Transform,and Hilbert-Huang Transform(HHT)that combines Empirical Mode Decomposition(EMD)and Hilbert Transform.The advantages and disadvantages of these methods are illustrated by analyzing the example.Secondly,due to the non-linear and non-stationary characteristics of the subsynchronous signal which contains multiple oscillation modes,the Synchrosqueezing Wavelet Transforms(SWT),which is a time-frequency analysis method that combines the Wavelet Transform and the reallocation methods-Synchrosqueezing,is used to analyze the subsynchronous signal and identify the parameters of signal.In order to improve the anti modal aliasing performance of the method,the suitable mother wavelet is selected to act as the wavelet bases of SWT in this paper.Besides,Kmeans clustering is used to improve the performance of SWT,leading to the ability to determine the frequency domain for signal reconstruction can be provided.The data of the numerical simulation signal and the IEEE first benchmark model of subsynchronous oscillation show that the SWT method combined with Kmeans clustering can locate the oscillation frequency accurately for modal reconstruction,and obtain high accuracy identify the result of the signal parameters of subsynchronous with Hilbert Transform.It is verified that the anti modal aliasing performance of SWT is better than the EMD algorithm,and the proposed method is effective and superior.Finally,the Fourier-Based Synchrosqueezing Transform(FSST)is used to study the detection and analysis of the subsynchronous oscillation signal,which has better performance on anti modal aliasing compared to SWT and EMD.The DBSCAN-Kmeans fusion clustering,which solves the problem that Kmeans clustering need to decide the number of clusters in advance,is used to improve FSST for making FSST can locate and reconstruct the subsynchronous oscillation modes with identifying the number and frequency of it accurately.The data of the numerical simulation signal and the IEEE benchmark model of subsynchronous oscillation show that FSST can separate the adjacent modes with closer distance,and obtain the parameter results with high accuracy through parameter identification.It is proved that FSST has better performance on signal analysis than SWT and the feasibility of the FSST method combined with DBSCAN-Kmeans fusion clustering in the detection and analysis of subsynchronous oscillation.