Effects of Duplication Operator in Evolutionary Simultaneous Design Optimization of Multiple Cars

We propose a duplication operator to increase the number of common design parts in the simultaneous design optimization of multiple cars. In the optimization problem, we optimize a combined sets of design variables of multiple cars simultaneously and tries to increase the number of common parts among multiple cars. In order to increase the number of common parts, the design variable values of the same design parts among multiple cars should be matched. However, it is a hard task to evolutionary algorithms treating each variable independently and varying it stochastically. To improve the number of common parts among multiple cars, we propose a duplication operator copying a variable value of a car to others, and verify its effectiveness when we combined with three algorithms, NSGA-III, MOEA/D, and TNSDM-A. As the results, we show that the proposed duplication operator improves the number of common parts and the combination with TNSDM-A achieves the highest search performance.