Author: Rafael David Tordecilla Madera
Programme: Doctoral Programme in Network and Information Technologies
Idioma: English
Supervision: Dr Àngel A. Juan Pérez & Dr Javier Panadero Martínez
Faculty / Institute: Doctoral School UOC
Subjects: Computer Science
Key words: biased-randomized heuristics, Monte Carlo simulation, simheuristics, fuzzy logic, routing problems, location problems
Area of knowledge: Network and Information Technologies
Summary
Transportation and logistics (T&L) are currently highly relevant functions in any competitive industry. Locating facilities or distributing goods to hundreds or thousands of customers are activities with a high degree of complexity, regardless of whether facilities and customers are placed all over the globe or in the same city. A countless number of alternative strategic, tactical, and operational decisions can be made in T&L systems; hence, reaching an optimal solution – e.g., a solution with the minimum cost or the maximum profit – is a really difficult challenge, even with the most powerful of computers. Approximate methods, such as heuristics, metaheuristics, and simheuristics, may be used to solve T&L problems. They do not guarantee optimal results, but they yield good solutions in short computational times. These characteristics become even more important when considering uncertainty conditions, since they increase T&L problems’ complexity. Modeling uncertainty involves the introduction of complex mathematical formulas and procedures, however, the model realism increases and, therefore, so does its reliability in representing real world situations. Stochastic approaches, which require the use of probability distributions, are among the approaches employed most often to model uncertain parameters. Alternatively, if the real world does not provide enough information to reliably estimate a probability distribution, then fuzzy logic approaches become an alternative to model uncertainty. Hence, the main objective of this thesis is to design hybrid algorithms that combine fuzzy and stochastic simulation with approximate and exact methods to solve T&L problems considering operational, tactical, and strategic decision levels. Therefore, biased-randomized heuristics and metaheuristics are firstly explained to solve T&L problems that only include deterministic parameters. Later, Monte Carlo simulation is introduced to these approaches to deal with stochastic parameters. Finally, fuzzy simheuristics are employed to address simultaneously fuzzy and stochastic uncertainty.