The use of neural networks for the automated management of crop moisture

Abstract

The highest yield of agricultural crops is achieved with the optimal amount of moisture, nutrition, heat, air and light. At the same time, the water regime of the soil is necessary for crops and is created by the appropriate irrigation regime, which establishes the norms, timing and amount of irrigation depending on the biological characteristics of the crops, natural and economic conditions. In determining the irrigation water flow, water consumption or total evaporation is taken into account, depending on climatic conditions, the amount of thermal energy supplied to the surface, soil moisture, the type and yield of the crop.

Therefore, issues of adaptation and self-study of automated soil moisture management systems under the influence of random weather factors, changes in the characteristics of the control object, improvement of control accuracy due to the operational accounting of the effect of disturbances on the object, ensuring the production of planned crop yields while rational use of energy and water resources are relevant.

In addition, modern moisture management systems for agricultural crops should not only ensure sufficient control accuracy, but also predict the plants need for water for a certain period, minimize energy and water costs without yield loss, be reliable and convenient in operation, provide the operator with complete and timely information about the value of all parameters and the state of the control system.

To solve these problems, an approach to automating the process of controlling irrigation systems using neural networks has been considered. The proposed approach allows to improve the accuracy of soil moisture management, to ensure the production of planned crop yields, to save water and energy resources due to their rational use.