Estimation of Model Order for Low Frequency Oscillations Utilizing Modified ICSS

Small signal disturbance is one of the problems in present power system network and the effects of the small signal disturbances is the poorly damped low frequency oscillations in power system network. There is a need of detection of these model order to maintain the stability. This paper focuses on a mathematical model based on the sudden change in the variance in sequence using modified integrated cumulative sum of square technique which involves a filtering model based on singular value decomposition and then performing the validation condition using K-medoids clustering to determine the model order which helps in estimation of the lower order frequencies and the damping parameter. It is being demonstrated that the proposed algorithm works even for lower value of signal to noise ratio superimposed in input signals and different PMU reporting rates. This algorithm outperforms in estimating the precise number of model order present in the signal. The robustness and accuracy of the proposed scheme is analyzed on a practical probing real time signals obtained from WECC network and the simulated synthetic signals including the white Gaussian noise. The comparison is performed with several methods for identification of modes at different noise level. Finally, the modes are being analyzed using the rotational in-variance technique based modified TLS-ESPRIT method.