Sistema inteligente de diagnóstico termográfico asistido por inteligencia artificial para la detección temprana de anomalías en equipos de subestaciones eléctricas de Guayaquil

Abstract

In order to detect thermal anomalies in Guayaquil's electrical substation equipment early on, this thesis focuses on developing an intelligent thermographic diagnostic system that is supported by artificial intelligence approaches. Predictive maintenance plans must be strengthened to guarantee the continuity and dependability of the electrical service due to the region's environmental conditions and the ongoing rise in energy consumption, which puts these facilities under increased operating demands. The study is based on the examination of thermographic pictures taken from important substation equipment, including main medium-voltage disconnectors, power transformers, and 69 kV GCB-type circuit breakers. As a non-invasive monitoring technique, infrared thermography makes it possible to assess the equipment's thermal behavior while it is operating normally without compromising service delivery. Based on the captured images, an artificial intelligence model is designed and trained using deep learning algorithms, particularly convolutional neural networks. This model aims to automatically identify and classify thermal anomalies, considering their severity level. The implementation of this automated diagnostic system seeks to reduce reliance on manual analysis, provide greater consistency in results, and facilitate decision-making in maintenance planning. The expected results of this study contribute to improved maintenance management in electrical substations, reducing the probability of unexpected failures, extending equipment lifespan, and strengthening operational safety. Furthermore, this work contributes to the development of applied knowledge in the field of electrical engineering by integrating infrared thermography, computer vision, and artificial intelligence, in line with current Industry 4.0 trends.