Sistema de información climática para caracterizar de forma aproximada el comportamiento de 5 variables mediante una técnica de aprendizaje automático en el área de influencia de la Universidad Distrital Francisco José De Caldas Facultad Tecnológica

Abstract

The document provides a detailed analysis of an innovative weather station that combines environmental data collection with advanced IoT technology and machine learning to predict weather conditions in the influence area of the Francisco José de Caldas District University, Technological Faculty. It presents a system that not only captures real-time data but also utilizes machine learning algorithms to interpret and forecast weather events. This interdisciplinary project demonstrates an integration of telecommunications engineering and control and automation engineering, showcasing the feasibility of low-cost and highly efficient weather stations for informed decision-making in various sectors such as agriculture, disaster management, and urban planning. The project stands out for its sustainability-focused design, employing devices such as the ESP32 microcontroller, known for its processing power and low energy consumption, Hall effect sensors to measure variables such as wind speed and direction, as well as precipitation. Google's Firebase platform plays a significant role in storing and analyzing the collected data, enabling real-time access and processing that are crucial for meteorological accuracy. Additionally, ESP-NOW technology is used to create cohesive and reliable networks between the sensors and the monitoring center. The project is justified not only for its educational and scientific value but also for its potential to improve everyday life and environmental management within the university community and its surroundings. With a futuristic vision, the document explores future innovations and trends, including deeper integration of IoT and machine learning, which could further transform weather analysis and prediction accuracy. The methodology includes the development of a remotely connected prototype weather station to a database and the use of preprocessing techniques and exploratory data analysis to predict atmospheric phenomena. Three machine learning algorithms are compared using statistical indices such as MSE, MAE, and efficiency index, and a detailed analysis of the collected data is provided to understand the relationships between different climatological variables such as temperature, humidity, wind direction and speed, and precipitation. The work emphasizes the importance of compliance with technical and quality standards and competence in the development and operation of weather stations, concluding with a discussion on the relevance of research for technological development and innovation in the region.