Research In Multiobjective Co-evolutionary Algorithm Based On Operator Hybridization

In the field of science and engineering,there are many multi-objective optimization problems and they are more complex generally.It is difficult to solve such problems by using the conventional operational research methods.However,heuristic-based intelligence algorithm has advantages of solving these optimization problems.The intelligent optimization algorithm is a population-based stochastic optimization algorithm which is inspired by “the survival of the fittest” theory.It simulates the evolution process or swarm intelligence behavior of the organisms in the natural environment.It can get a set of solutions each time which improves the algorithm's efficiency.Therefore,the intelligent evolutionary algorithm receives more and more attention and a variety of evolutionary algorithms have been proposed.Comparing with single-population optimization,co-evolution algorithms which benefit from operator hybridization and population collaboration,have shown promising performances in tackling multi-objective optimization problems.This thesis mainly studies the relevant intelligent optimization algorithms from the perspective of complementary advantages and analyzes the characteristics of various algorithms.At last,this thesis proposes a co-evolutionary algorithm which can combine the advantages of different types of algorithms and increases the application of evolutionary algorithms.The main work of this thesis is as follows:(1)This thesis presents a novel PSO-DE co-evolutionary algorithm based on decomposition framework(MODEPSO).The algorithm employs three populations and integrates particle swarm optimization(PSO),simulated binary crosses(SBX)and differential evolution(DE)into the framework of decomposition.The elite solutions got by PSO and SBX are considered as evolutionary candidates which will be further evolved by DE operation.The coherence between different populations enhances the versatility and robustness of the algorithm.(2)This thesis presents a more general bi-population based co-evolution framework(BCF)and realizes a bi-population based multiobjective co-evolutionary algorithm with adaptive operator hybridization(BMCA)as a representative.The BMCA uses an adaptive operator hybridization approach and a resource allocation strategy.The former adjusts the proportion of different operators by monitoring the average performance of different evolution operators.And the latter allocates evolutionary resources according to the evolutionary state of different subproblems.The offspring from each evolution operator can update the two populations simultaneously and exchange information.This can improve the convergence rate and population diversity.