| Summary: | Real-world complex optimization problems are one of the most complex challenges faced by scientific community. Achieving the best solution for a complex problem in an acceptable time interval is not always possible. In order to solve this problem, metaheuristics are one of the available resources. Having this in mind, finding a technique among others that presents better results in most executions would allow solution choosing to be more directive and assertive. Most used techniques comprise metaheuristics. These allow to find an acceptable solution in an acceptable time interval, even if the achieved solution was not the optimal possible. In this sense, this thesis intends to analyse four optimization techniques. Two population based techniques, one of them based in the behaviour of the bees in colony (Bee Colony) and another based in computational evolution (Genetic Algorithms). And, two single solution techniques, one based in memory structures (Tabu Search) and another based in the metallurgy industry (Simulated Annealing). These techniques were applied to two different optimization problems and computational results were registered and analysed. A prototype was built and used to obtain the results of applying metaheuristics to the Travelling Salesman problem (TSP) and the Knapsack Problem (KP). Evaluating the results, it was not possible to prove either that all algorithms are equivalent or that one of them is better in the majority of the cases.
|
|---|