Novel approaches for online modal estimation of power systems using PMUs data contaminated with outliers

One of the most important issues in modal estimation of power systems using PMUs data is the negative effect of outliers. Hence, in addition to the techniques of analyzing PMUs data, the necessity of implementing some kinds of approach to overcome these outliers is tangible. This paper aims to present different approaches to overcome outliers and also estimate the electromechanical modes of the system accurately when there is suspicion that the PMUs data may be contaminated by discordant measurements. Proposed approaches are generally categorized into two main classifications: the first category detects and modifies outliers in the pre-processing stage adaptively and then prepares the modified data for processing. The second category consists of robust approaches that detect the outliers through processing data and solving the problem. Based on adaptive data pre-processing with a different data modification method, another approach is proposed in this research for cases where parametric techniques are employed for modal estimation. The efficiency of the proposed methods is investigated on the sixteen-machine five-area test system. It is assumed that all sets of simulated data are distorted by outliers in a few samples. Finally, the results are compared and the application of each method is discussed.