Solving Multiple-Container Packing Problems using Pseudo-Meiosis Genetic Algorithm

Abstract

Knapsack problems are a class of common but difficult (NP-complete or NP – hard) problems. Since, it is believed that no knapsack problem algorithm can be constructed whose computation time optimality increases as any polynomial function of the problem size. There is a variety of knapsack-type problems in which a set of entities, together with their values (profits)and sizes, is given, and it is desired to select one or more disjoint subsets so that the total of the sizes in each subset does not exceed given bounds and the total of the selected values is maximized .Diploid representation and dominance operator are advanced operators that attempt to improve upon the power of traditional genetic algorithms .Pseudo – Meiosis Genetic Algorithm(PsM GA) is one form of genetic algorithms that incorporate diploidy structure and dominance mechanism in their genetic search .the goal of this dissertation is to present the application of PsM GA in one of the promising combinatorial optimization problems- the Knapsack Problem (KP).Results obtained concern two types of KP: the 0/1 KP and the Multiple Container Packing Problem, MCPP. Moreover, several aspects are considered in experiments such as , the algorithm used for evaluation of the individuals (fitness evaluation ), the number of items (i.e., search space size ), the correlation between the weights and the profits of items, and the capacity of the knapsack.