ResearcherID - CLICK HERE Scientific Journal Impact Factor (SJIF-2020) - CLICK HERE

NEUROCOMPUTING AND ITS APPLICATION IN THE SUGAR INDUSTRY

Vasiliy Tkachenko, Elena Popova, Marina Vorobeva, Daniil Zhmurko

The article is devoted to the development and adaptation of the tools of neuro-spectral analysis for the performance indicators of large agro-industrial holdings of the sugar sub-complex of the agro-industrial complex. Currently, neurocomputing is often used to analyze data, and therefore it is appropriate to compare it with old, well-developed statistical methods. In the authors' literature review on statistics (econometrics), it is often stated that the use of neurocomputing is an ineffective tool for analyzing the main components, regression and discriminant models. It is also noted that multilayer neural networks can actually solve problems such as regression and classification. However, firstly, the processing of data by neural networks is much more diverse, for example, the active classification by Hopfield networks or Kohonen feature maps, which have no statistical analogs. Secondly, many studies concerning the application of the combined approach in the form of neural networks and spectral analysis in the agricultural economy have revealed their advantages over classical statistical methods. The necessity of a predictive assessment of the performance of large agro-industrial holdings of the sugar sub-complex of the agro-industrial complex is substantiated. Such forecast estimates can be obtained on the basis of the implementation of a neuro-spectral analysis describing the strategic management of large-scale enterprises of the sugar sub-complex along the optimal development trajectory. It is proved that the identified cycles make it possible to predict with a high probability the main trends in both regional and global data on the production of sugar and sugar beet (or cane). The authors proposed the use of adapted tools in the form of neuro-spectral analysis in solving problems of predicting the performance of integrated production systems of the sugar sub-complex of the agro-industrial complex. Neuro-spectral analysis allows to improve the quality of the forecast for the development of complex dynamic processes to a greater extent than the classical spectral analysis in its pure form. This makes it possible to continue to develop strategically adjusted integrated management decisions. The results of the solution of the formulated problem were obtained and analyzed by the method of neuro-spectral analysis. The results of its application in the tasks of forecasting have demonstrated the possibility of solving them and confirmed their practical significance in the field of management of large industry enterprises of the sugar subcomplex. In parallel with this, the problem of forecasting high-frequency oscillations has been solved. Key words: neurocomputing, Fourier transform, neurospectral analysis, prediction, cycles, frequencies.

FULL TEXT