Multi-objective Fruit Fly Optimization Based on Cloud Model

Abstract

A normal cloud model based fruit fly optimization algorithm is extended for multi-objective optimization. The normal cloud model is used to produce new food positions for the fruit flies in the osphresis-based search, and a parameter self-adaptive strategy is proposed to adjust the search space dynamically. The concept of Pareto domination is incorporated into the vision-based search and a cloud model based multi-objective fruit fly optimization algorithm (CMMOFOA) is then developed. To have a good diversity of the member in the elitist archive, a normalized nearest neighbor distance based diversity maintenance strategy is adopted. The UF test set in CEC2009 is used to test the performance of CMMOFOA. Numerical simulation results demonstrate that CMMOFOA can converge to the Pareto fronts of the test problems with good accuracy and distribution, and has better or competitive performance compared with six popular state-of-the-art algorithms.