Abstract

In view of the shortcomings of the usual genetic algorithm when solving multi-objective combinatorial optimization problems, the paper proposes an alternately evolving genetic algorithm. It adopts alternate strategy and optimizes multiple objectives one by one circularly. The solving results of the Sudoku puzzle indicate that this strategy can make the excellent patterns of different objectives all grow rapidly, and the results’ comparison verify its feasibility and excellence in the general convergence. The sensitivity of the algorithm’s parameter is also analyzed.