Creation of the spatial metric for the image of an agricultural object

The analysis of various approaches to forecasting complex multifactorial systems in conditions of uncertainty of external conditions is presented. It is necessary to develop these approaches in order to create adequate models of agricultural activities for their effective planning and management. A distinctive feature of agricultural production is a critical dependence on environmental factors, which cannot be accurately predicted. Regression modeling and analysis of time series used at present to solve this problem in difficult cases do not result in an adequate forecast of the dynamics of an agricultural object. As an approach, it is proposed to use the construction of the "image" of the system. This approach is classified as "nature-like", as it simulates a way of decision-making by a specialist on the basis of accumulated experience and intuition. The key parameter of this construction will be the correct choice of the metric (coordinate system). This approach is illustrated by an example of creating an image of a two-dimensional phenomenon in a one-dimensional coordinate system. As a result, an image is understood as an image of reality in a vector space of a certain dimension. The image in the authors' view is a reflection of reality in an artificially created metric, more suitable for understanding and analysis, but retaining the main (important) features and functions of the original object. Artificial intelligence techniques can be seen as tools for image creation and analysis. An important characteristic of an image is its predictive power, i.e. the ability to use the image in order to predict the state of a real object in the future period. An image retains its predictive power if the forecast obtained using this image corresponds to the data obtained when observing a real object. The image is formed in a suitable metric for solving a specific problem. The key metric parameter of the image of agricultural activity, suitable for forecasting purposes, is the minimum dimension of the vector space used, at which the predictive power of the image is retained to solve the problem.

1. Amosov H.M. Modeling of thinking and psyche. Kiev: Naukova dumka Publ., 1965, 303 p. (In Russian).

2. Lewandowsky S., Farrell S. Computational modeling in cognition: Principles and practice. SAGE publications, Inc., 2011, 359 r.

3. Shabrov N.N., Kurikov N.N. Analysis and visualization of scientific research results using virtual reality technologies. Nauchnotekhnicheskie vedomosti SPbPU. Estestvennye i inzhenernye nauki = St. Petersburg State Polytechnic University Journal of Engineering Science and Technology, 2011, no. 4 (135), pp. 200–205. (In Russian).

4. Shabrov N.N. Virtual environment software and hardware complexes are key components of virtual engineering technologies. CAD/CAM/ CAE Observer, 2016, no. 3 (103), pp. 83–86. (In Russian).

5. Ogorodnikov P.I., Usik V.V. Forecasting the production and yield of grain crops based on regression models. Vestnik Orenburgskogo gosudarstvennogo universiteta. = Vestnik Orenburg State University, 2011, no. 13 (132), pp. 354–359. (In Russian).

6. Zatonskii A.V., Sirotina N.A. Forecasting economic systems using a model based on a regression differential equation. Ekonomika i matematicheskie metody = Economics and Mathematical Methods, 2014, vol. 50, no. 1, pp. 91–99. (In Russian).

7. Adamadziev K.R., Kasimova T.M. Methods of forecasting of development of agriculture. Fundamental'nye issledovaniya = Fundamental Research, 2014, vol. 1, no. 5. (In Russian).

8. Burda A.G., Mokropulo A.A., Polusmak V.I., Burda S.A. Multicollinearity in rating models of evaluation of investment projects of agroeconomic systems. Fundamental'nye issledovaniya = Fundamental Research, 2019, no. 3, pp. 11–16. (In Russian).

9. Moiseev N.A. Methods for increasing the reliability of predictive econometric studies. M.: Ruscience Publ., 2019, 272 p. (In Russian).

10. Sall' M.A. Climate Risks: Temporal Trends and Heteroscedasticity. Meteorologiya i gidrologiya = Russian Meteorology and Hydrology, 2015, no. 7, pp. 84–92. (In Russian).

11. Istigecheva E.V., Mitsel' A.A. Models with autoregressive conditional heteroscedasticity. Doklady Tomskogo gosudarstvennogo universiteta sistem upravleniya i radioelektroniki = Proceedings of the TUSUR University, 2006, no. 5 (13), pp. 15–21. (In Russian).

12. Kantorovich G.G. Time series analysis. Ekonomicheskii zhurnal Vysshei shkoly ekonomiki = Higher School of Economics Economic Journal, 2002, vol. 6, no. 4, pp. 498–523. (In Russian).

13. Afanas'eva T.V. Modeling fuzzy time series trends. Ul'yanovsk: Izdatel'stvo Ul'yanovskogo gosudarstvennogo tekhnicheskogo universiteta = Publishing House of Ulyanvosk State Technical University, 2013, 215 p. (In Russian).

14. Tat'yankin V.M. Using multilayer neural networks in time series forecasting. Prioritetnye napravleniya razvitiya nauki i obrazovaniya = Priority directions of development of science and education, 2014, no. 3, pp. 195–197. (In Russian).

15. Solso R. Cognitive psychology. SPb.: Piter Publ., 2006, 589 p. (In Russian).

16. Emel'yanova Yu.G., Fralenko V.P. Methods of cognitive-graphical representation of information for effective monitoring of complex technical systems. Programmnye sistemy: teoriya i prilozheniya = Program Systems: Theory and Applications, 2018, vol. 9, no. 4 (39), pp. 117– 158. (In Russian).