An Approach to Maintain the Balance between Exploitation and Exploration of the Evolutionary Process in Multi-objective Algorithms

Abstract

Multi-objective optimization has been applied in many fields of science, including engineering, economics, finance, and logistics, where optimal decisions need to be taken in the presence of trade-offs between two or more conflicting objectives. There are several techniques to solve multi-objective optimization problems in which evolutionary algorithms are often used because they simulate the principle of natural evolution and work on population. In evolutionary algorithms, to ensure the ability to find solutions globally and quickly find the optimal solution, we must maintain the exploratory and exploitative capacities of the evolution, which also means the exploration and the exploitation of algorithms. In multi-objective optimization, population quality in diversity and convergence is essential to achieve the best possible solution set. The analysis showed that the relationship between the properties of the population directed by evolution and the ability to explore and exploit the evolutionary process is quite clear. This research evaluated the population quality according to generations of the evolutionary process based on popular measures and adjusted the algorithm to create an equilibrium transformation of those metrics, thereby better maintaining the balance between the exploration and exploitation of the population. Experiments performed on the direction-based multi-objective evolutionary algorithm with typical benchmark sets showed that the results bring good performance both in terms of solution quality and balance of the evolutionary process.