Utilizing factor models to simulate emergency shutdowns of power grids

Abstract

Abstract In the article, based on statistics on emergency outages in the Levoberezhniy and Pravoberezhniy districts of Irkutsk and the sums of average daily temperatures in years 2010-2017 regression models are built. As an auxiliary factor for the models built on values that were grouped by year, the seasonality index was calculated using the “Kassandra” model. In the model, built for values that were grouped by month, a trend was added in addition to the seasonal component. A retrospective prediction of emergency outages based on the data of the Levoberezhniy district, gathered in April 2020 showed that a model with a trend and a seasonal component gives a more accurate prediction. Since the sum of the average daily temperatures can be considered a random variable, the proposed factor models can be used to predict emergency outages only with some probability. The solution of inverse problems, when the factor value is determined for a given number of emergency outages, can be used to assess the risks associated with the influence of low and high temperatures on the objects of the electric grid complex.