Implementation of a multiple sensory system for the detection of fluid losses in ducts through adaptive neuro-fuzzy inference

• Autores: Carbajal-López, Marco, Valdivia-Diaz, Anthony, Briones-Zuñiga, Jose, Sotomayor-Beltran, Carlos
• Formato: artículo
• Fecha de Publicación: 2024
• Institución: Universidad Tecnológica del Perú
• Repositorio: UTP-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.utp.edu.pe:20.500.12867/14609
• Enlace del recurso: https://hdl.handle.net/20.500.12867/14609; https://doi.org/10.14445/22315381/IJETT-V72I11P128
• Nivel de acceso: acceso abierto
• Materia: Leak detection; Pressure; Neural network; TensorFlow; https://purl.org/pe-repo/ocde/ford#2.02.04

This work presents a sensor-based fluid pipeline leak detection system using an Adaptive Neuro-Fuzzy Inference System (ANFIS). The implemented model consists of multiple flow sensors and pressure differentials, processed through a hybrid neural network and fuzzy logic hybrid system. The network comprises five layers, with layer 1 being in charge of performing the membership function and layer 5 being part of the calculation of the exit rule. The system aims to detect and prevent fluid leakage in fluid passages by identifying changes in fluid flow and pressure differential. The results demonstrate that the ANFIS system can accurately detect leaks in the ducts, reaching 93.24 % of accuracy, indicating the percentage of correct predictions in the training set. Additionally, a validation set not part of the training was used for the model's generalization ability. This data set allowed us to measure patterns and characteristics of the model with new and previously unseen data.