FORECASTING THE GROUNDWATER LEVEL OF CEMENT RAW MATERIALS DEPOSIT BASED ON DYNAMIC NEIGHBORHOOD MODELS

Abstract

Introduction. The development of a mathematical model for the groundwater level of a deposit of cement raw materials located in the Zadonian-Yelets aquifer, which is the principal domestic water supply source for the city of Lipetsk, is considered. Therefore, it is necessary to provide ongoing monitoring and to have the possibility to predict the water level under the field development. The work objectives are the identification and study of a dynamic neighborhood model with variable hierarchical neighborhoods of the groundwater level that enables to adequately predict value of the water level in the examined wells.Materials and Methods. The definition of a dynamic neighborhood model with variable hierarchical neighborhoods is given, differing by time-varying double-level neighborhood communications between the first- and second-level nodes. At each next discrete instant of time, the neighborhood model nodes change their state under the influence of the online parameters and node states included in their neighborhood. As a subcase, we consider a model with line state recalculation functions. Parametric identification of the dynamic neighborhood model consists in finding the system parameters for each second-level node, and is based on the ordinary least squares.Research Results. A linear dynamic neighborhood model with variable hierarchical neighborhoods for predicting the groundwater level in a cement raw material deposit located in the Zadonian-Yelets aquifer is developed. The software using C++ is developed for the parametric identification and simulation of the functioning of the dynamic neighborhood model under consideration. It enables to determine parameters of the node state recalculation functions for a given structure, and also to predict the model behavior in the operation process. A hierarchical structure is given, and a parametric identification of the linear dynamic neighborhood model of the groundwater level is carried out. After the parametric identification on the teaching data selection, the mathematical model is checked on the test sample.Discussion and Conclusions. The obtained average ratio errors of the identification and forecast suggest the developed model validity and enable to recommend it for predicting the underground water level of a cement raw materials deposit.