Research On Large-Scale Sparse Multi-Objective Optimization Problem And Application Based On Membrane Evolutionary Algorithm

As a new computing model,membrane computing is widely used in various fields because its great parallelism and uncertainty of rule execution,such as combination optimization and engineering planning.Membrane evolutionary algorithm is an evolutionary algorithm designed based on membrane computing,which abstracts the life activities such as cell division,merger and death and the information exchange between cells.Membrane evolutionary algorithm uses the parallel optimization of membrane computing and heuristic information to achieve individual parallel evolution,which shows great advantages in global search ability.Multi-objective optimization problem is an extremely widely used topic in optimization problems.The purpose of its solution is to seek a compromise method,so that each objective to be optimized achieves a balance that satisfies the decision maker.As a powerful method of combinatorial problems,evolutionary algorithm is applied to multi-objective optimization problem,and exhibits good performance.In recent years,with the in-depth study of large-scale multi-objective optimization problems,some multi-objective optimization evolutionary algorithms have been unable to meet the needs of this rapid increase in search space.Therefore,it is of great significance to design algorithms suitable for large-scale sparse multi-objective optimization problems.This thesis starts with the large-scale sparse multi-objective optimization problem,improves the shortcomings of some existing algorithms,and proposes a multi-objective optimization evolutionary algorithm framework MOEA/IDVE(An Evolutionary Algorithm based on Initialization Decision Variables Evaluation).Further,combined with the idea of membrane computing,a membrane evolutionary algorithm is designed to solve the critical node detection problem,named MEA/CNDP(A Membrane Evolutionary Algorithm for Solving Biobjective Critical Node Detection Problem).The main work of this thesis is as follows:(1)In view of the large-scale sparse multi-objective optimization problems,a population initialization strategy based on decision variable evaluation is proposed,so that each decision variable can be more accurately evaluated in the population initialization stage.Moreover,the relative multi-objective optimization evolutionary algorithm framework MOEA/IDVE is proposed.Finally,the effectiveness of the proposed algorithm is verified by comparative experiments.(2)As an application of large-scale sparse multi-objective optimization problem,this thesis studies the bio-objective optimization model of the critical node detection problem.Based on MOEA/IDVE,Combining the idea and characteristics of membrane evolutionary algorithm,MEA-CNDP is proposed to solve the critical node detection problem.MEA-CNDP improves the population initialization strategy of MOEA/IDVE,thereby designing a complete membrane evolutionary algorithm framework and evolutionary operators,and adding auxiliary strategies to improve the performance of the algorithm.Finally,based on the data set of the critical node detection problems in practical application,the relevant experiments are designed.The experimental results verify the effectiveness of MEA-CNDP to solve such problems.In this thesis,the membrane computing is effectively combined with the multi-objective optimization problem.This thesis proposes effective solution for the large-scale sparse multi-objective optimization problem,and extends it to the application of critical node detection problem based on membrane evolutionary algorithm.This is not only a new attempt,but also has a certain reference value for scientific research in related fields.