Impact of seasonal weather on forecasting of power quality disturbances in distribution grids

Power supply disruptions, including short-time disturbances, can lead to large direct and indirect financial losses. The ability to predict the risk of these disturbances allows for preventive actions and increases the reliability of the supply. This paper investigates the impact of using seasonal data of combined common weather conditions on the power quality prediction in distribution grids. Our main contribution consists of weather-based predictive models for three types of events that frequently occur in these grids, as well as an analysis of the influence of two training approaches: with either seasonal or all-year data, on their performance. All developed models score higher than arbitrary guessing; in several instances the improvement is considerable. It is demonstrated that in some cases the models improve when the training data is limited to a subset corresponding to a particular meteorological season. Examining variable importance values and distributions of the models' data, it is shown that this situation takes place particularly when weather conditions correlated with the occurrence of power grid events vary across seasons.