Optimizing particle swarm optimization to solve knapsack problem

Abstract

Knapsack problem, a typical problem of combinatorial optimization in operational research, has broad applied foregrounds. This paper applies particle swarm optimization to solve discrete 0/1 knapsack problem. However, traditional particle swarm optimization has nonnegligible disadvantages: all the parameters in the formula affect the abilities of local searching and global searching greatly, which is liable to converge too early and fall into the situation of local optimum. This paper modifies traditional particle swarm optimization, and makes the position of particle which achieves global optimization reinitializated. Through analyzing the final result, the paper has proven that the improved algorithm could improve searching ability of particle swarm, avoid converging too early and solve 0/1 knapsack problem more effectively. © Springer-Verlag 2010.