Propuesta de gestión para la detección de fallas en colectores de aguas residuales del distrito de San Juan De Lurigancho

Abstract

In this thesis, three failure scenarios were evaluated in wastewater collectors of the San Juan de Lurigancho district. The first analysis consisted in evaluating the sewer’s condition according to the diameter, the slope, the time (in months), the number of failures and the depth, and the second analysis consisted in evaluating if a collector suffers a stuck according to the number of activities using like variables to the diameter, the slope, the speed, the tie, the flow, the angle, the wet perimeter, the area, time (in months) and the hydraulic diameter. The analysis was performed using the method known as Machine Learning through the clustering method, what served to do the predictions in both scenarios through k-means clustering method then apply the 4 regressions which are the logistic, the K-NN, the SVM and the Decision Tree to know, thought of their confusion matrix, where the accuracy ratios and the error ratios were calculated, which of the four was the most correct for each scenario.

Date of Award	2021
Original language	Spanish (Peru)
Supervisor	Elmer Ramirez Quiroz (Asesor)