Abstract:

Aiming at solving difficulties from the complexity of optimization model and the mixture of variable types in the practical problems, an ant colony optimization algorithm is proposed that can sufficiently deal with mixed continuous and ordinal or categorical discrete by constructing the pheromone model and designing ant probabilistic solution construction approaches for mixed variables. Then a mixed variable and any time oriented ant colony optimization algorithm is generated by further considering the unknown number of objective function evolutions and expensive scenario in the practical problems, the any time optimization oriented evaluation indicator is designed for algorithm parameters and automatically configure algorithm for improving solution quality and optimization efficiency. Finally, the generated algorithm is tested on various engineering optimization benchmark problems. Compared with results from the literature, the proposed algorithm demonstrates its effectiveness and robustness.