Research On Multi-objective Particle Swarm Optimization Algorithm With Information Feedback

In real life,multi-objective optimization problems(MOPs)include many problems.MOPs include multiple objectives that need to be optimized simultaneously,Traditional methods are difficult to solve MOPs.In order to better solve MOPs,researchers have conducted in-depth research on the optimization algorithm.Algorithms can get a set of better solutions without considering the character of objective functions,which play a significant role to deal with MOPs.Among them,multi-objective particle swarm optimization(MOPSO)performs well on MOPs.The findings indicate that the feedback of individual historical useful information to the later optimization process can improve the performance of the algorithm.However,for MOPs,the traditional information feedback method is difficult to be applied directly.In order to apply the feedback information directly to MOPs and improve the performance of the algorithm,three new information feedback methods suitable for MOPs are proposed in this paper and combined with MOPSO.The main research contents are as follows:(1)A multi-objective particle swarm optimization algorithm based on mutual information feedback model(MOPSO-MIF)is proposed in order to apply the feedback information directly to MOPs.This method determines the weight of each individual in the information feedback model by calculating the mutual information value between individuals.This method can directly apply the information feedback model to multi-objective optimization problems and ensure that the optimized particles have a certain historical inheritance.Ten test functions and two real optimization problems are used to test the effectiveness of the algorithm.(2)A multi-objective particle swarm optimization algorithm based on fuzzy inference feedback model(MOPSO-MEOP)is proposed in order to reduce the influence of the calculation of mutual information value between individuals in the mutual information feedback model on the calculation cost of the algorithm.The algorithm combines the fuzzy inference strategy based on the maximum entropy principle with the one-step prediction strategy to predict the weight of each individual in the information feedback model,The running time of the algorithm is reduced by this model.The update strategy of competitive particle swarm optimization algorithm (CSO)is used to update the individuals in the external archive,The convergence speed of the algorithm is improved by the update strategy.Twenty test functions and multi threshold image segmentation problems are used to test the effectiveness of the algorithm.(3)A multi-objective particle swarm optimization algorithm based on velocity feedback model(V-MOPSO)is proposed in order to solve the problem,which is difficult to determine the parameters of information feedback model for MOPs.This method uses the speed between individuals to determine the weight of individuals in the feedback model.In the early stage,the individual search efficiency is improved by this method;In the later stage,the population diversity of the algorithm is improved by this method.Comparative experiments with the other five MOPSO are carried out on five test functions,and the algorithm is used to optimize two real optimization problems to verify the feasibility of the improved algorithm.To sum up,this paper comprehensively studies and analyzes the traditional MOPSO,puts forward three effective improved algorithms and applies them to practical optimization problems,which provides a new method for improving the performance of MOPSO.