Early Warning of Low-Frequency Oscillations in Power System Using Rough Set and Cloud Model

Abstract

The stability of the power system is largely affected by low-frequency oscillations, so early warning research on low-frequency oscillations in power grids has become an urgent task. Traditional low-frequency oscillation early warning methods are still deficient in handling incomplete and highly discrete information. Compared with the existing methods, we have pioneered a synergistic mechanism of discrete attribute screening and continuous probabilistic feature fusion by combining the dynamic attribute approximation algorithm of rough sets with the cloud model, which effectively solves the loss of information caused by the discretization of continuous data in the traditional methods. Firstly, we analyze the principle of grid oscillation, use rough sets to process the raw data and indicators, remove redundant attributes, and get the set reflecting the relationship of different attributes. Then we construct a standard cloud based on grid operation data and a comprehensive cloud based on PMU data and obtain the oscillation warning evaluation. Finally, through the validation and simulation of 10 machine and 39 node systems in New England, as well as the comparison with other methods, the rationality and effectiveness of the proposed method are proved to be of theoretical and practical application value.