Automatización y modelado predictivo para el tratamiento de datos de las estaciones de carreteras en Castilla y León
Loading...
Official URL
Full text at PDC
Publication date
2025
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
Este proyecto se enmarca en la iniciativa Territorio Rural Inteligente de Castilla y León y tiene como objetivo la mejora de la gestión de incidencias y la predicción de condiciones adversas en las carreteras mediante el uso de tecnologías IoT, automatización y modelos de inteligencia artificial.
En una primera fase, se desarrollaron diversos flujos de trabajo automatizados que permite detectar fallos en los sensores desplegados en las estaciones de carretera y generar incidencias automáticamente a través de la plataforma EasyVista. Esto evita la revisión manual de más de 2000 entidades, optimizando significativamente el proceso.
En una segunda fase, se abordó el desarrollo de un modelo de IA para predecir condiciones adversas. Para ello, se realizó un análisis previo de las tendencias históricas de los parámetros recogidos, prestando especial atención a la preparación y limpieza de los datos, lo cual fue clave para la fiabilidad del modelo. Se probaron diferentes algoritmos de machine learning, y se compararon sus métricas de rendimiento con el fin
de seleccionar el modelo más eficiente y preciso.
Finalmente, se implementó un sistema que permite enviar informes por correo electrónico, incluyendo tanto las predicciones generadas como información relevante sobre el estado de las estaciones de sensores.
This project is part of the Smart Rural Territory initiative of Castilla y León and aims to improve incident management and the prediction of adverse road conditions through the use of IoT technologies, automation, and artificial intelligence models. In the first phase, various automated workflows were developed to detect failures in the sensors deployed at road stations and to automatically generate incidents through the EasyVista platform. This eliminates the need for manual review of over 2,000 entities, significantly optimizing the process. In the second phase, the development of an AI model to predict adverse conditions was undertaken. For this, a preliminary analysis of historical parameter trends was carried out, with particular attention paid to data preparation and cleaning, which was key to the model's reliability. Different machine learning algorithms were tested, and their performance metrics were compared to select the most efficient and accurate model. Finally, a system was implemented to send email reports, including both the generated predictions and relevant information about the status of the sensor stations.
This project is part of the Smart Rural Territory initiative of Castilla y León and aims to improve incident management and the prediction of adverse road conditions through the use of IoT technologies, automation, and artificial intelligence models. In the first phase, various automated workflows were developed to detect failures in the sensors deployed at road stations and to automatically generate incidents through the EasyVista platform. This eliminates the need for manual review of over 2,000 entities, significantly optimizing the process. In the second phase, the development of an AI model to predict adverse conditions was undertaken. For this, a preliminary analysis of historical parameter trends was carried out, with particular attention paid to data preparation and cleaning, which was key to the model's reliability. Different machine learning algorithms were tested, and their performance metrics were compared to select the most efficient and accurate model. Finally, a system was implemented to send email reports, including both the generated predictions and relevant information about the status of the sensor stations.
Description
Trabajo de Fin de Máster en Internet de las Cosas, Facultad de Informática UCM, Departamento de Arquitectura de Computadores y Automática, Curso 2025/2026













