A Sparrow Search Algorithm Based On Multi-Strategy Fusion And Its Application

There are optimization problems everywhere in real life.People have many methods to solve optimization problems.One of the most common and effective methods is to use swarm intelligence algorithm.The inspiration of swarm intelligence algorithm comes from the group behavior of natural creatures.People put forward many mathematical models to solve the optimal solution of optimization problems by studying the rules of division and cooperation of creatures in the population.Because swarm intelligence algorithm has the advantages of insensitivity to the advantages and disadvantages of initial solution selection,no requirements for fitness function,and certain extensibility,it is widely used to deal with practical problems and has important practical significance.This article mainly studies the sparrow search algorithm.By analyzing the shortcomings of the mathematical model of the sparrow search algorithm,two improved sparrow search algorithms are proposed.At the same time,benchmark test functions are used to conduct multiple comparative experiments for verification.The specific research content is as follows:(1)A sparrow search algorithm(POMSSA)based on oscillation and somersault strategies was proposed,Introducing a flip factor to enhance the algorithm’s local optimization ability;Introducing polynomial mutation perturbation strategy to improve the algorithm’s ability to jump out of local optima.Subsequently,this article tested the performance of the improved sparrow search algorithm through three sets of experiments: ablation experiments were conducted on 10 standard test functions to verify the effectiveness of the POMSSA algorithm improvement strategy;Conduct comparative experiments between different optimization functions,highlighting the excellent performance of the POMSSA algorithm compared to the classic SSA algorithm,classic GSA algorithm,and classic GWO algorithm;Finally,Wilcoxon rank sum test was performed to verify the significant differences in the improved strategies.(2)A Sparrow Search Algorithm(TNRSSA)based on adaptive normal distribution movement and random disturbance is proposed.The TNRSSA algorithm generates an initial population through Tent chaos,increasing population diversity;Improve global search by introducing an adaptive step size factor;Introduce randomly perturbed individuals,improve the update formula,and enhance the algorithm’s ability to jump out of local optima by increasing the weight impact of random individuals in the population.This article tests the performance of the improved sparrow search algorithm through three sets of experiments: conducting comparative experiments between different optimization functions on 10 standard test functions,which are different from the classic SSA algorithm,classic GSA algorithm,and classic GWO algorithm,highlighting the performance advantages and differences of the POMSSA algorithm;Compare it with two newly proposed sparrow search algorithms to further verify the superiority of the improved strategy.Finally,perform Wilcoxon rank sum test to verify the significant difference of the improved strategy.(3)Apply two improved sparrow search algorithms to the K-means remote sensing image segmentation algorithm.Due to the fact that the advantages and disadvantages of the classic K-means algorithm rely on the selection of initial clustering centers,which are randomly generated in the classic K-means algorithm,it affects the stability of the algorithm and is prone to errors.In this paper,The optimal value found by the algorithm is used as the initial clustering center of K-means clustering algorithm.In order to improve the algorithm performance,the effectiveness of the improved K-means was verified through experimental comparison with several improved K-means algorithms in terms of visual effects and segmentation accuracy.