Scientific Journal of KubSAU

Improvement of consistency and effectiveness of strategic planning and forecasting in modern conditions requires the development of the existing classifications of types of planning, strategies, forecasts and forecast methods. This work examines the introduction to problems of spectral analysis of the macroeconomic dynamics of worldwide and Russian sugar subcomplex. The article is devoted to forecasting the activities of integrated manufacturing systems of sugar subcomplex in agro industrial subcomplex. The article deals with aspects of practical application of economic-mathematical methods (based on spectral analysis) to control the economic parameters of the integrated industrial systems of the sugar subcomplex, oriented to meet the needs of the sugar production of the people not only of separate regions, but also of the country as a whole. Procedures of identifying and research of periodic components of the dynamics of the development of the agriculture segment are based on methods of spectral analysis of random processes. The study describes the performed experiments with various kinds of non-stationary time series of agricultural sector and food industry sugar subcomplex. The article provides examples of the results of numerical experiments the spectra of time series of sugar production, sown areas, gross harvest and yield of sugar beet and sugar cane across countries, systematic ideas and methods underlying the spectral analysis. The estimation of obtained results is given in article. The author’s algorithm for the adaptive method of spectral analysis was implemented in the context of a specific software product, named MS Excel. The results of the empirical research confirmed the possibility of practical use of developed models in forecasting possible scenarios for the development of sugar subcomplex in the interests of integrated production systems. The results are illustrated by numerous graphs based on real data. The projections of latent structures of sugar subcomplex by macroregions are built. It is revealed that each of the macroeconomic time series can contain at least from 2 to 13 harmonics (cycles) of different kind and strength of impact on the trend

The article considers brief theoretical information of the wavelet transform and the methods of identification of nonlinear time-varying systems using multi resolution wavelet transform. The methods of data processing based on wavelet transformation are widely used in recent times. Wavelets have significant advantages compared to Fourier transform because wavelet transform tells you about not only the frequency spectrum of the signal, but also on what point in time came one or another harmonic. With their help, you can easily analyze intermittent signals or signals with powerful bursts. Moreover, wavelets allow us to analyze data according to scale, on one of the preset levels (small or large). The unique properties of wavelets allow constructing a basis in which the representation of the data will be expressed with just a few nonzero coefficients. This property makes wavelets a useful tool for data packaging. Small expansion coefficients may be discarded in accordance with the selected algorithm without a significant impact on the quality of the compressed data. Wavelets have found wide application in digital signal processing and data analysis. There are two classes of wavelet transforms: continuous and discrete. In the article we have implemented the discrete wavelet transform with the resulting output distribution on a 3D graph. The algorithm and the results of converting a time series of indicators of integrated industrial systems of sugar subcomplex in agro industrial subcomplex. The methods of neural network modeling for improved accuracy in predicting high-frequency oscillation are applied in the research. The method of determination of cyclic patterns based on coefficients of the wavelet transform has been proposed

The article considers brief theoretical information of the wavelet transform and the methods of identification of nonlinear time-varying systems using multiresolution wavelet transform. The methods of data processing based on wavelet transformation are widely used in recent times. Wavelets have significant advantages compared to Fourier transform because wavelet transform tells you about not only the frequency spectrum of the signal, but also on what point in time came one or another harmonic. With their help, you can easily analyze intermittent signals or signals with powerful bursts. Moreover, wavelets allow us to analyze data according to scale, on one of the preset levels (small or large). The unique properties of wavelets allow constructing a basis in which the representation of the data will be expressed with just a few nonzero coefficients. This property makes wavelets a useful tool for data packaging. Small expansion coefficients may be discarded in accordance with the selected algorithm without a significant impact on the quality of the compressed data. Wavelets have found wide application in digital signal processing and data analysis. There are two classes of wavelet transforms: continuous and discrete. In the article implemented the discrete wavelet transform with the resulting output distribution on a 3D graph. The algorithm and the results of converting a time series of indicators of integrated industrial systems of the sugar subcomplex in the agro industrial subcomplex. The methods of neural network modeling for improved accuracy in predicting high-frequency oscillation are applied in the research. The method of determination of cyclic patterns based on coefficients of the wavelet transform is proposed