Optimisation par l'algorithme des rorquals application à des benchmarks en mécanique

Abstract

The Whale Optimization Algorithm (WOA) is an optimization algorithm recently developed by Mirjalili and Lewis in 2016. It is a metaheuristic algorithm using a population of random solutions proposed for problem solving of optimization. The proposed algorithm solves continuous variable optimization problems. In this paper, this algorithm has been modified to deal with mixedvariable problems. After that, it was applied to the optimization of two mechanical devices under functional stress. The first device, a tension/compression spring in this case, was considered to validate the program developed for the WOA algorithm. The second where it is a question of a bolted plate coupling is the project considered in the context of this Master 2 thesis. For the first device, i. e. the compression tension spring, it is a question of minimizing the weight of the latter under unequal constraints. The problem depends on three decision variables, the diameter of the wire which is a normalized variable, the number of turns which is an integer variable and the external diameter of the turns which is a continuous variable. The second problem with a bolted plate coupling has a discrete variable, an integer variable and two continuous variables. The problem is subject to three constraints of inequality and depends on five discrete parameters, which depend on the choice of bolt diameter. The program developed for the WOA algorithm, under Matlab environment, was validated and applied to the project successfully. The results achieved are superior to those presented by authors who have considered other metaheuristic and have shown that the programmes carried out lead to optimal solutions with complete satisfaction of functional constraints