Research On Dynamic Multi-objective Optimization Method Based On Social Learning Optimization Algorithm

At present,many complex problems in industrial production and scientific computing can be transformed into dynamic multi-objective optimization problems.Conducting in-depth research on these problems and proposing effective solution methods have important application value.In recent years,domestic and foreign scholars have conducted in-depth research on dynamic multi-objective optimization problems and proposed a series of effective methods.However,existing methods do not achieve ideal results when complex environments undergo drastic changes.In addition,in real life,decision-makers are usually only interested in a portion of the solutions,rather than all the solutions.At this point,continuing to seek all solutions will result in wastage of computing resources.In response to this issue,some researchers have proposed preference aware dynamic multi-objective optimization problems and provided some effective solving algorithms.However,the types of preference changes in these solving algorithms are relatively fixed,which greatly limits the generality of the algorithms.Therefore,how to design a solution algorithm that meets the different types of preference changes of decision-makers remains an urgent and challenging problem to be solved.In order to efficiently solve dynamic multi-objective optimization problems in complex and dynamic environments,as well as dynamic multi-objective optimization problems with uncertain preference changes,this article delves into the following research work:(1)In order to efficiently solve dynamic multi-objective optimization problems in complex environments,this paper proposes a dynamic multi-objective optimization method based on hybrid prediction strategies and improved social learning optimization algorithms.Specifically,when the environment changes,the method first generates a portion of the population based on a representative individual prediction strategy;Secondly,generate another part of new groups based on the inflection point prediction strategy;In particular,in order to improve the diversity of the population,a certain number of new individuals are randomly generated based on the historical information and environmental changes of the algorithm iteration.Furthermore,in order to improve the efficiency of problem solving,this paper improves the social learning optimization algorithm,and designs operators suitable for dynamic multi-objective optimization problems for each evolutionary space;Finally,a new dynamic multi-objective optimization method is formed by combining a hybrid prediction strategy with an improved social learning optimization algorithm.In this paper,FDA,d MOP,and F function sets are used as experimental test function sets to compare their performance with four representative algorithms;The performance of this method is analyzed in depth using Inverse Generation Distance(IGD).Experimental results show that the proposed method has good convergence and robustness.(2)Aiming at the problem of uncertain decision-maker preference types in dynamic multi-objective optimization problems,this paper proposes a dynamic multi-objective optimization method based on preference perception and improved social learning optimization algorithm.Specifically,this method first considers two preference changes and proposes a preference perception strategy;When preferences change over time and environment,the preference history information is obtained through the least squares fitting method to obtain the preference movement trajectory,and the movement direction and step size of the next preference change are predicted.This method can quickly find the optimal solution when preferences change;When preference information changes and environmental changes are independent of each other,the inflection point will evolve towards the preference point direction,and some individuals will be randomly generated around the preference point to ensure population diversity while leading the evolution of the population;In addition,a dynamic multi-objective optimization method based on preference perception and improved social learning optimization algorithms is formed by combining the social learning optimization algorithm with the proposed preference dynamic strategy and r dominance relationship.Finally,a large number of validation experiments were conducted based on various testing problems,and the experimental results showed that the convergence speed and stability of this method were better than other comparative algorithms.