Solving 01 Knapsack Problem with variations of Flower Pollination Algorithm

Abstract

The 01 knapsack problem is a combinatorial optimization problem where objective is to maximum profit. It can be considered as a resource allocation problem where the decision is to pick the most important items. Knapsack problem has a large range of applications in many domains. This paper presents three variations of FPA influenced by concept of crossover and mutation operator in genetic algorithms. The three variations, named FPA with crossover and mutation (Version 1), FPA with crossover (Version 2), FPA with mutation (Version 3) for solving 01 knapsack problem are presented. Comparison of variations is done with basic FPA algorithms and other algorithms for three datasets of a single dimension of 01 knapsack problem. The results show that improved FPA has better results than basic FPA. The results are optimal for smaller instances and with increase in number of objects results are closer to optimal but not optimal.