RDF description Dr. Pedro López

Postdoctoral Research Assistant

Sep. 2013  -  Present


social-network-icon LinkedIn

Aplicación de Técnicas de Inteligencia Artificial para la Predicción de Congestiones a Corto Plazo

Directed by: Enrique Onieva     Co-advisor:  Asier Perallos

 02 Dec 2016 - 17:00
 University of Deusto
 Cum Laude by unanimity

149 pages (≈ 1.8 MB)

Intelligent Transportation Systems (ITS) can be defined as the intersection between communication and electronic technologies with the aim of mitigating transports problems, whatever their type is. ITSs are applied to areas such as vehicular communications, traffic signal control systems, or road information systems. ITSs are really important in road transport, whether for personal trips, or transport of goods or people. One of the main problems of road transport is the traffic congestion. Its early detection could help to take actions to decrease the noise, not only in urban streets but also in freeways; reduce the amount of polluting gases; to increase the performance and accuracy of road transport systems, and to save in public infrastructure for institutions. For all of this, the early detection of traffic congestion is a fundamental topic in ITS research field. Along the last decade, and with the goal of giving a solution to this problem, autoregressive techniques, vehicular communication methods, and Soft Computing techniques have been applied. Focusing on the last ones, fuzzy logic and metaheuristics have emerged as good alternatives in the last years thanks to their adaptability and the easy way to represent the solutions they offer. In this context, a hybrid method that combines several metaheuristics has been developed in order to apply it to the optimization of fuzzy logic techniques for congestion forecasting. This algorithm counts with the ability of adapt itself to different problems, modifying its operators and internal configuration. On the one hand, to validate the formulated hypothesis, traffic datasets have been created with real data. As a result, it is possible to evaluate the different configurations of the algorithm and its application to fuzzy systems. Also, the configurations are compared with themselves as well as with well-known techniques of the literature. With this comparative, the good performance of the proposed technique is proved. On the other hand, mathematical functions of different types and complexity have been used as benchmarks to prove the adaptability of the technique to other themes. Besides, an in-deep study has been carried out in order to choose the parameters of the technique when it is applied to functions optimization.