An Effective SOM-Based Approach for low-Quality PMU Measures to Recognize Oscillatory Modes in Power System

To sustain the reliability of the connected grid system’s performance, it’s crucial to spot any low frequency oscillations that are poorly damped. But in actual, due to the increase in the installment of Phasor measurement units (PMUs), the data collected at Phasor data concentrators (PDCs) may contain non - linear data with missing values and outliers, which results in a decay in the performance of the estimator and affects system stability. This can happen due to heavy communication network congestion, malicious attacks, malfunctioning PMUs or PDCs, among other factors. In this situation, this paper discusses the use of Self organizing maps (SOM) technique in the beginning to give a substantial data set to handle such a problem. The second phase involves identifying the oscillatory modes using the improved total least square estimation ofjsignal parameters through rotational invariance technique (TLS-ESPRIT) algorithm. For generated test signals with missing data and outliers at various noise levels, the suggested method has been compared against Bayesian low-rank matrix/Hankel tensor factorization (BLMF/BLTF) with k-Medoids clustering and ML-based ERA (ML-ERA). The effectiveness of the suggested algorithm is further illustrated on two area data and real probing data collected from Western Electricity Co-ordinating Council (WECC).