Simulation of neural-network control temperature and humidity conditions in the greenhouse

На основі нейронних мереж та з використанням експериментальних даних створено модель зміни температурно-вологісного режиму у теплиці в залежності від потужностей обігрівачів ґрунту та повітря, що дозволило забезпечити формування енергоефективних навчальних вибірок в адаптивній системі керування мікрокліматом. The purpose of this article is to study the temperature and humidity in the greenhouse mode: assessment of the importance of the impact of input parameters for the weekend using a neural network model developed model test and confirm its suitability for use in simulation and in the practical implementation. Based on neural networks (type - radial basis functions) and the experimental data using a model change in temperature and humidity regime in a greenhouse to obtain reaction facility management on control actions. Check the adequacy of this model proved its suitability for use in simulation. The use of this neural network is practically possible to ensure the formation of energy efficiency training samples for later use in adaptive (intelligent) system controls temperature and humidity conditions in the greenhouse. The results of simulation modeling intellectual functioning control system of temperature and humidity conditions in the greenhouse revealed that its qualitative characteristics of the following (for both control outputs): overshoot: the upper front - 0%, the lower front - 2.5 - 3.75%; the maximum dynamic deflection: the upper front - 0 Bottom front - 0.02 - 0.03; napivkolyvan number: the upper front - 0 Bottom front - one that meets the technical requirements