In the car industry, companies want to use the smallest number of different kinds of components in a series of cars and retain high design freedom of shared components. This thesis will quantify these two goals and look at this problem, named commonality optimization, from two perspectives: One directly as a bi-objective optimization and one as a bi-level optimization. For the former, Rook model is used to encode the configuration. Then Iterative Elitist Genetic Algorithm (IEGA), Niched Elitist Genetic Algorithm (NEGA) and Niched Pareto Genetic Algorithm-II (NPGA-II) are introduced for solving the problem. As to the latter, this thesis focus on the lower level of the bi-level optimization. For it, dynamic objective genetic algorithm (DOGA) and Monte Carlo Tree Search(MCTS) method is used after mapping the optimization problem into a deepest node search problem. Those algorithms are tested in two small cases and observed that NSGA-II and MCTS with the Hasse diagram as its underlying model
(MCTS-HD) are the most suitable algorithms for industrial use. In the end, the result of implementing them in a real case is shown, in which it is impossible to enumerate all possible solutions to find the optimum.
«In the car industry, companies want to use the smallest number of different kinds of components in a series of cars and retain high design freedom of shared components. This thesis will quantify these two goals and look at this problem, named commonality optimization, from two perspectives: One directly as a bi-objective optimization and one as a bi-level optimization. For the former, Rook model is used to encode the configuration. Then Iterative Elitist Genetic Algorithm (IEGA), Niched Elitist...
»