Margarita N. Barsukova, Sofia A. Petrova, Yaroslav M. Ivanyo

Irkutsk State Agricultural University named after A.A. Ezhevsky,

In the article, the results of modeling of long standing time series with using the method of multilevel division of sequences were presented. The upper level of the selected series reflects favorable conditions of obtaining products, the lower one characterizes an unfavorable situation, and the intermediate one indicates a certain averaged version of the activity of an agricultural enterprise. At same time, linear and nonlinear trends with saturation level were used, which make it possible to predict indicators of biological productivity of agricultural crops in favorable, unfavorable and average conditions for medium and long term perspectives. The proposed multilevel trends in the parametric model of optimizing of production of agricultural products were applied. The models on real objects were implemented. These models can be used to management agricultural production

optimization, multilevel model, trend, forecasting, agricultural production, management