CC BY
1
УДК 519.246.78
РЕШЕНИЕ ЗАДАЧ ПРОГНОЗИРОВАНИЯ С ПРИМЕНЕНИЕМ НЕЙРОННЫХ
СЕТЕЙ ДЛЯ ЦЕНТРОВ ОПЕРАТИВНОГО УПРАВЛЕНИЯ РЕГИОНОМ
*
И.А. Панфилов, Т.В. Пен
Сибирский государственный университет науки и технологий имени академика М. Ф. Решетнева Российская Федерация, 660037, г. Красноярск, просп. им. газ. «Красноярский рабочий», 31
E-mail: pen_tatiana@mail.ru
В представленной работе применение нейронных сетей в задачах прогнозирования временных рядов.
Ключевые слова: статистический анализ, нейронные сети, многослойный персептрон.
SOLUTION OF FORECASTING PROBLEMS USING NEURAL NETWORKS FOR REGIONAL OPERATIONAL CONTROL CENTERS
I.A Panfilov, T V. Pen*
Reshetnev Siberian State University of Science and Technology 31, Krasnoyarskii rabochii prospekt, Krasnoyarsk, 660037, Russian Federation
E-mail: pen_tatiana@mail.ru
In the presented work, the application of neural networks in the tasks offorecasting time series.
Keywords: statistical analysis, neural networks, industrial production index, multilayer perceptron.
The 21st century is considered the age of information technology. Indeed, if you look around, it is difficult to imagine that once there were not all digital devices, gadgets, standard computing techniques. But time does not stand still, the world develops, and society develops with it. The economy is reaching a new level, new economic indicators appear for assessing and making managerial decisions in adopting a strategy for managing the country.
The digitalization of society allows you to move to a completely new level of decision-making -not based on rumors, news or subjective opinions of experts, but on the basis of data, and this is a key change that accompanies digitalization.
The main directions in digitalization are the creation of operational control centers, centers for receiving and processing citizens' appeals. Thanks to such approaches, in the future, on the basis of the obtained values, it becomes possible to analyze the received data, generate statistics, classify data, forecast statistical indicators and make strategic decisions within the framework of state governance.
Operational control centers are necessary for timely response to emergencies, to determine the scale of the problem, to find vulnerabilities and to predict the course of events for a certain period of time. Thanks to digitalization, the necessary data is collected into databases. The amount of data is too large and manual processing is almost impossible; for this, mathematical models for solving the necessary problem are created. To solve the necessary problems, a mathematical apparatus is most often used - neural networks. Since the algorithm is self-learning, it is enough to set the necessary parameters and configure the network. Further, the model will improve itself and the error will be minimized.
Секция «Математические методы моделирования, управления и анализа данных»
The foundation for digitalization is the development of information and analytical systems that can download and visualize data, adapt to the data customer's requirements, which will allow making forecasts and making management decisions, displaying analytical conclusions with any discreteness.
For example, in the Krasnoyarsk, a «Regional Management Center» was formed, which creates a system where various statistical data from all departments of the region are co-opted. The data is automatically displayed in the form of dashboards, which is a convenient tool for monitoring and developing industries.
Consider forecasting time series using a statistical and structural model, then we will analyze the results obtained and compare them, and also determine the model that is most effective for constructing predicted values, and then making management decisions. To compare the results obtained, we will use the statistics of the birth rates of organizations from January 2016. to December 2020 [1].
The next year will be the predicted year for analyzing and comparing the accuracy of the constructed models [2].
The backpropagation rule of the network is based on gradient descent. During the training period, the weights are adjusted to reduce the error [3].
Completion of training allows you to optimize synoptic weighted connections as well as establish reliable connections between inputs and outputs.
Table 1
Received values
Forecasted values for 2021 Original series Regression equation Forecast error Neural network Forecast error
January 8,2 9,99 0,291 7,03 0,123
February 8,3 9,96 0,251 7,7 0,032
March 10,2 9,99 0,004 8,58 0,236
April 11,4 9,88 0,210 8,88 0,575
May 8 9,83 0,304 8,81 0,060
June 8,8 9,80 0,091 8,39 0,015
July 8,7 9,83 0,116 9,01 0,009
August 7,8 9,72 0,335 8,64 0,066
September 8,1 9,67 0,224 8,96 0,069
October 9,3 9,64 0,011 9,22 0,000
November 8,2 9,67 0,196 9,17 0,087
December 9 9,56 0,029 8,6 0,014
Amount 2,061 1,286
Fig. 1. Forecasting the fertility rate of organizations for 2021
Based on the results obtained, it can be concluded that the predicted indicators of the neural network have a greater approximation to the initial values of the series than the indicators obtained by the regression equation. The neural network provides a more adequate model for describing the economic indicators of a time series, while the regression equation averages the values.
Based on the results obtained, using a mathematical apparatus - a neural network, a forecast of the birth rate of organizations for 2021 was built.
Based on the graph, we can conclude that the trend is decreasing, but in the future the indicator will stabilize. The approximation error when training the network was 9.5%. This indicates that this model is suitable for predicting the fertility rate of organizations.
Thus, we can conclude that digitalization is an integral part of the modern world. Thanks to it, the best control and management in various areas of the state is carried out.
References
1. State statistics [Electronic resource] - 2021. - URL: http: // gks.ru/ (accessed date: 03/01/2021)
2. Khaikin S. Neural networks: full course, 2nd edition; trans. from English - M .: Publishing Williams House, 2006.
3. Kremer N.Sh., Prutko B.A. Econometrics / Textbook for high schools. - M .: UNITY-DANA, 2005.
© Panfilov I. A., Pen T. V., 2021
