Abstract:

The wolf pack algorithm has been proposed based on inspiration by group survival swarm intelligence of the wolf pack, and successfully applied to complex function optimization problems and the normal 0-1 knapsack problem. To solve the multidimensional knapsack problem (MKP), a tryingloading repair operator based on the MKP specific knowledge is designed to effectively repair and improve infeasible solutions. Then, traditional objective function based on large penalty parameters has been improved so as to reduce the risk of easily trapping in the local optima. Meanwhile, inspired by the reproductive rule of the wolf pack, an improved binary wolf pack algorithm (IBWPA) is proposed for solving the MKP. Simulation results based on 19 benchmark MKP instances with different scales and comparative analysis between the IBWPA and other algorithms demonstrate the effectiveness and computational robustness of the proposed algorithm.