Research On The Improvement And Application Of Two Swarm Intelligence Algorithms

Optimization problems are a popular research topic in various fields,and they are challenging due to their complexity and practical application requirements.Traditional methods usually have poor performance and efficiency when dealing with functions that are discontinuous or non-differentiable.Additionally,in practical engineering,most optimization problems have many local solutions,and some high-dimensional problems have a large search space,making it difficult to find the optimal solution.Swarm intelligence algorithms for large-scale,complicated,nonlinear optimization problems are a promising method.They are highly flexible,robust,and efficient.With the development of technology,these algorithms have gained widespread attention from scholars.They can effectively solve problems without requiring specific information and are suitable for many practical applications.Therefore,swarm intelligence optimization algorithms have become an important research direction for solving optimization problems.After years of research,various swarm intelligence algorithms have emerged,among which the Gray Wolf Optimization Algorithm(GWO)and the Sparrow Search Algorithm(SSA)are new intelligent optimization algorithms proposed in the past decade.These two algorithms have the advantages of a simple structure,few control parameters,good precision,and minimal algorithm complexity when compared to other algorithms.However,research has shown that these two algorithms suffer from the problems of uneven initial populations,imbalanced local development and global exploration capabilities,and being trapped in local optima,which to some extent,affects their performance.Therefore,to improve the above shortcomings,this paper aims to improve the Gray Wolf Optimization Algorithm and the Sparrow Search Algorithm,making them more competitive compared to existing optimization and improvement algorithms and can be successfully applied to some field problems,with significant research significance and practical value.The main work of this paper is as follows:(1)A multi-strategy improved gray wolf optimization algorithm(MIGWO)was presented to address the problem of gray wolf optimization algorithm with unequal beginning population,unbalanced local development and global exploration capabilities,and slipping into local optimum.The algorithm’s convergence speed and solution accuracy are improved first by initializing the gray wolf population using the reverse learning strategy.Second,a nonlinear convergence factor is introduced to adjust the control parameters to further balance GWO’s global search and local development capabilities.Then,the cuckoo optimization algorithm’s search mechanism is paired with the position update process to assist the gray wolf population in breaking out of the local optimum when it becomes stagnant and to improve the algorithm’s global optimization potential.MIGWO outperforms original GWO and other enhanced GWO in simulation trials on 9 benchmark functions.(2)An improved sparrow search algorithm(Multi-strategy enhanced sparrow search algorithm,or MESSA)is proposed in order to address the issue that the initial population of the sparrow search algorithm is uneven,the global search ability is weak,and it is simple to fall into the local optimum in the later stage of iteration.First,the Bernoulli map is used to start the population to assure population variety and uniformity,as well as to increase the algorithm’s convergence speed.Second,the Levy flight is used to boost the global search ability of individual finders in the sparrow population.The cross-optimization algorithm is then introduced to assist the sparrow.When the population is static,remove the stagnant state,exit the local optimum,and increase solution accuracy.It is demonstrated through 9 benchmark function studies that MESSA can effectively increase the convergence speed and solution accuracy of SSA,as well as jump out of the local optimum.(3)MIGWO and MESSA can be used to solve traveling salesman and feature selection problems,respectively.When the traveling salesman problem is solved using The experimental findings demonstrate that MIGWO’s search performance is superior due to its quick convergence time,robust global optimization search ability,and capacity to exit local optimal solutions.That demonstrates the utility of MIGWO.MESSA is used in conjunction with the sigmoid conversion function to create a binarybased enhanced sparrow search algorithm(Binary enhanced sparrow search algorithm,BESSA)for feature selection.The results of the experiment on 10 UCI classification data sets and one real world data set demonstrate that BESSA is capable of effectively choosing the appropriate feature subset to raise classification accuracy.