Research On Hybrid Multi-Objective Evolutionary Algorithm Based On Decomposition And Its Application

In the field of scientific research and engineering,many problems are composed of multiple conflicting goals.We call this problem as a multi-objective optimization problem.The population-based evolutionary algorithm can obtain an approximate Pareto solution set in a single run,so the multi-objective evolutionary algorithm has became a more general and effective method to solve the multi-objective optimization problem.In recent years,the multi-objective evolutionary algorithm based on decomposition has been used in many multi-objective evolutionary algorithms.It uses a decomposition method to decompose a multi-objective optimization problem into multiple single-objective optimization sub-problems and optimize them at the same time.However,the MOEA/D algorithm has many shortcomings such as loss of diversity and slow convergence when solving multi-objective optimization problems.In this paper,we improve the MOEA/D algorithm in three aspects,such as the generation of weight vector,the decomposition method of sub problems and the generation operator of offspring,and propose an improved algorithm named Hybrid-MOEA/D-ASD.At the same time,according to the characteristics of wireless sensor networks,two algorithms are proposed based on the MOEA/D framework,Hybrid-MOEA/D-? and Hybrid-MOEA/D-?,which are used to solve the multi-objective coverage optimization problem in wireless sensor networks.The main work of this paper is as follows:(1)Aiming at the improvement of MOEA/D algorithm,Hybrid Multi-objective Evolutionary Algorithm Based on Decomposition with Adaptive Search Direction(Hybrid-MOEA/D-ASD)is proposed.The algorithm uses a multi-objective evolutio-nary algorithm based on decomposition as a framework,In the method of weight vector generation,the search direction is controlled by the weight vector of adaptive adjustment sub-problem;In the decomposition strategy,the Tchebycheff Approach and the sub-population decomposition method are adopted;The advantages of different crossover operators are fully utilized by the hybrid crossover strategy in the generation generation operator.Experimental results show,Compared with the classical multi-objective evolutionary algorithms MOEA/D,MOEA/D-M2M and MOEA/D-DRA,The algorithm proposed in this paper Hybrid-MOEA/D-ASD has better convergence and diversity in the ZDT and DTLZ series of standard test functions.(2)Two optimization goals of coverage and network lifetime,the optimization algorithm of wireless sensor network based on decomposition is considered.Because the energy balance of the whole network has a great influence on the performance of the network,the objective function of energy consumption equilibrium is added on the basis of the algorithm.In order to solve the algorithm using the genetic algorithm single search strategy to bring the limitations of the search results,the hybrid-MOEA/D-? algorithm is proposed based on the classical MOEA/D algorithm,using the cross-mixing method of genetic algorithm and differential algorithm to optimize each sub-problem.Because Hybrid-MOEA/D-? lacks the preservation of high quality individuals and the optimal solution of the individual concentration of very few.In this paper,an improved discrete binary particle swarm optimization algorithm is proposed,and a hybrid-MOEA/D-? algorithm is proposed.Since the weight of each sub-problem in the Hybrid-MOEA/D-? algorithm is fixed,the search direction is also determined to some extent.In order to further optimize each sub-problem,attempting to change the search direction is about the Hybrid-MOEA/D-? output as an improved binary particle swarm initial solution.The improved binary particle swarm algorithm is used to optimize the solution of each sub-problem so that the solution is better.The experimental results show that the hybrid-MOEA/D-? algorithm extends the network lifetime compared with other algorithms.