MODELING AND FORECASTING OF PETROLEUM PRODUCTS PRODUCTION TAKING INTO ACCOUNT SEASONALITY ON THE BASIS OF AUTOREGRESSIVE MODELS

Abstract

The relevance.Motorgasoline, diesel fuel and other oil products differ in seasonality of consumption, a way of transportation, the territory of realization. The change in demand for petroleum products under the influence of these factors leads to a change in the range and volume of production of various fuels. In this regard, the actual task is to plan the production of petroleum products. It is proposed to solve the problem of predicting the production of petroleum products using autoregressive models taking into account the seasonality factor. The main aimof the research is todevelopand select mathematical models suitable for forecasting generation and consumption of petroleum products production. Objects: production of petroleum products. The models are based on the data of a single interdepartmental information and statistical system. Methods are based on the use of methods of mathematical and simulation modeling. Results. The authors have carried out the review of methods of modeling time series of production and consumption of fuel and energy resources.It compares various mathematical models for forecasting the production of motor gasoline through the example of the Volga federal district. The authors developed the models that differ in consideration of the seasonal components and the type of trend. The use of a multiplicative model containing a trend in the form of a linear, autoregressive, autoregressive-power model with the calculation of the seasonality index is proposed. The articles shows that the best results as per the criterion of the average relative error were obtained using the model with autoregressive-power trend. The efficiency of the obtained model is shown on the example of federal districts of the Russian Federation to assess the production of motor gasoline and diesel fuel. The research results were obtained using the Matlab software package. The authors made post-forecast production of fuels according to the proposed model with an average relative error not exceeding 11 %.

Keywords

• petroleum product
• multiplicative model
• production
• time series
• seasonality index
• quality indicators.
• quality indicators