Análise das pressões do sistema de abastecimento de água do CCTA-UFCG utilizando simulação hidráulica e Random Forest.

Abstract

Water scarcity is a global issue, with projections indicating that many countries may face a severe water crisis. One alternative that seeks to assist in decision-making in order to contribute to water management is risk analysis, which aims to identify, evaluate and understand the possible problems that the system may face in the future. This work was developed with the objective of carrying out a risk analysis of collapse of the CCTA-UFCG water supply system due to variations in pressures varying over time, using hydraulic simulation and Random Forest. To meet the purpose described, a series of methodological procedures were necessary, such as the preparation of consumption data and network location, simulations with three different scenarios (175, 190 and 205), machine learning and the application of the Random Forest algorithm. The results demonstrated high performance of the Random Forest model, with accuracies of up to 100% in several target nodes, such as TO-16, Block 2 and Block 3, even under different pressure scenarios. The analysis of the importance of the variables revealed that nodes such as TO-18, Teachers' Block and TO-14 were recurrent among the most influential, indicating their criticality in the system. This integrated approach allowed us to accurately identify the most sensitive points in the network, favoring preventive actions and greater efficiency in the water management of the campus.