A K-Medoids-LSTM Based Technique for Electromechanical Modes Identification for Synchrophasor Applications

In this proposed work an efficient K-Medoids-LSTM based technique that takes into account of the degraded Phasor measurement unit (PMU) data for the estimation of poorly damped modes for wide area monitoring in smart grid is presented. This technique is designed in such a way that the detrimental effect of data missing and outliers which are created due to congestion in communication network, malfunction of PMUs or Phasor data concentrators (PDCs), and malicious attacks on mode estimation are mitigated. Here, the detection and removal of outliers are treated by applying K-Medoid algorithm, thereby the Long Short-term Memory (LSTM) is exploited for missing data imputation. Finally, total-least square-estimation-of-signal-parameters via rotational invariance technique (TLS-ESPRIT) is applied for mode estimation. The effectiveness and robustness of the proposed approach is validated by conducting statistical analysis study on synthetic signal through Monte Carlo simulation and compared with other recently developed techniques. This technique is also validated on Two area data and real probing data obtained from Western Electricity Co-ordinating Council (WECC).