Despite the intensive development of computer systems being introduced at the coal enterprises to provide air and gas monitoring, security is still not high enough, so that emergencies continue to occur due to a high concentration of explosive gases. Therefore, development of methods of forecasting the content of combustible gases in mines, which are used to improve the quality of air and gas assessment of the situation is urgent.In order to solve the problem of forecasting in the article the most common methods of forecasting – extrapolation, mathematical, associative, were analyzed.On the basis of the comparative characteristics the choice was made in favor of the neural network method. The main criteria for the choice of a particular neural network were such as the presence of feedback, the delay in the input layer, fast learning and prediction accuracy. Amongst all the networks that meet the criteria, the most suitable one is distributed TLFN.In order to determine the selected network architecture some numerical experiments were carried out. The criterion for the selection of architecture was the minimum MSE. According to the results, network architecture with the number of neurons to 10 of the study was chosen.In order to evaluate the effectiveness of the proposed method numerical studies that prove the effectiveness of the selected network architecture and its learning algorithm were carried out.

Рассмотрены и проанализированы существующие методы прогноза содержания метана. Исходя из основных преимуществ и недостатков, разработан и реализован нейросетевой способ прогноза, в основу которого положена распределенная сеть прямого распространения с задержкой во времени (distributed TLFN). Архитектура определена на основе проведенных экспериментов. Критерием выбора архитектуры было минимальное значение MSE.

Розглянуто та проаналізовано існуючі методи прогнозу вмісту метану. Виходячи з основних переваг і недоліків, розроблено і реалізовано нейромережевий спосіб прогнозу, в основу якого покладено розподілену мережу прямого розповсюдження з затримкою в часі (distributed TLFN). Архітектура визначена на основі проведених експериментів. Критерієм вибору архітектури було мінімальне значення MSE.