Abstract:

An estimation of core distribution method to approximately estimate the evolution population and the Pareto optimization solution set for the problem in the next environment was given. When a change in the environment was detected, the algorithm used the collected information of solutions from the previous difficult environments to predict the evolution population and the Pareto optimization solution set in the next environment. Consequently, the searching effectiveness of the proposed algorithm can be greatly improved. By the reasonable designing of evolution operators, a new dynamic multiobjective optimization evolutionary algorithm based on estimation of core distribution was proposed. Simulations were made on four widely used benchmark problems, and the results indicated that the proposed algorithm was very effective.

Key words: dynamic multi-objective optimization, evolutionary algorithm, estimation of distribution, Pareto optimization solutions