Abstract:

The wolf pack algorithm (WPA), inspired by swarm intelligence of wolf pack in their prey hunting behaviors and distribution mode, has been proposed and successfully applied in complex function optimization problems. Based on the designing of the move operator, the artificial wolves’ position, step-length and intelligent behaviors are redesigned by binary coding, and a binary wolf pack algorithm (BWPA) is proposed to solve combinatorial optimization problems in discrete spaces. BWPA preserves the feature of cooperative searching based on job distribution of the wolf pack and is applied to 10 classic 0-1 knapsack problems. Moreover, the 3 high-dimensional 0-1 knapsack problems are tested. All results show that BWPA has better global convergence and computational robustness and outperforms the binary particle swarm optimization algorithm, the greedy genetic algorithm and the quantum genetic algorithm, especially for high-dimensional knapsack problems.