Method of Gas Consumption Change-point Detection Based on Seasonally Multicomponent Model

A multi-component change-point model was used to take into account season changes in city gas consumption. On the basis of this model the method for determining the time series of gas consumption has been developed. The method is based on the use of the “Caterpillar-SSA” numerical method to separate the components of the model, followed by the random component analysis modified by the Brodsky-Darhovsky method. The obtained time moments of change-point make it possible to separate the annual time series into seasons, which will improve the accuracy of gas consumption prediction due to the smaller variance of the season segment.