An Analysis of Particle Swarm Optimization of Multi-objective Knapsack Problem

Abstract

Based on the Particle Swarm Optimization (PSO), two methods are adopted to solve the multi-objective knapsack problem, which is a knapsack problem with multiple objective functions. One is the weighted method: each objective function is given a certain weight according to the actual demand, so the multi-objective problem is transformed into an one objective problem and then PSO is used to solve it. Another one is the method of non-inferior solution: PSO is combined with the non-inferior solution. Therefore, more feasible solutions are shown in the final result, and the decision-maker can select the best solution according to the performance of values. The experiment shows that to solve the multi-objective knapsack problem, the non-inferior method is better than the weighted method because the former is more convenient for decision-makers and does not depend on the weights, which is important for the latter and may confuse the decision-maker.