APPLICATION OF NEURONETWORK TECHNOLOGIES FOR THE IDENTIFICATION OF THE MOISTURE CONDUCTIVITY COEFFICIENT OF THE SOIL

Abstract

Information on moisture conductivity can be used for mathematical (quantitative) analysis of various cases of soil moisture transfer occurring in natural conditions. This includes soil water filtration, water flow from the water table to the surface, soil moisture uptake, and soil moisture flow to plant roots. Water potential is a universal function that reflects the influence of all factors that can affect the energy state of water in the soil. As you know, water always moves from a higher potential to a lower one. The active force is the free energy or soil water potential gradient that represents the force that causes isothermal water flow. In isothermal conditions, the components of the water potential are capillary pressure, osmotic forces, and gravity. As a result, the moisture conductivity of unsaturated soils is not characterized by a single value (as in the case of saturated water transport), but rather is a function of suction pressure or soil moisture. The complex nature of the dependence of soil and water in terms of "moisture - potential energy - conductive moisture" is often presented in the form of empirical formulas and graphs. Many empirical formulas have been proposed to approximate the moisture conductivity function, which reproduce this function at certain time intervals with a certain accuracy. Currently, the task of developing technologies that ensure the economic and ecological efficiency of water regulation in drainage and humidification systems is relevant. In this regard, it is necessary to create a complete mathematical model of the soil based on the moisture transfer equation, one of the main parameters of which is the moisture conductivity coefficient. The purpose of the article is to develop a new method for determining the water conductivity parameter of the unsaturated zone of the soil based on a direct multilayer static artificial neural network, which contributes to increasing the accuracy of the measurement. The coefficient of moisture conductivity, which describes the physical processes in the soil, can be determined using a number of empirical formulas that contain empirical coefficients. An alternative method is the use of a neural network, with the help of which the coefficient of moisture conductivity or any other parameter of the soil, depending on the experimental data, is set to a sufficiently high value based on the studied sample. precision. A neural network trained on the training data set can be successfully used on independent test samples for a specific soil type that is not included in the training set.