Research And Implementation Of Improved Fruit Fly Algorithm Optimized Support Vector Machine Regression

With the development of industrial production,scientific research and other fields,traditional optimization algorithms have been unable to handle high-latitude,strongconstraints,and high-complexity optimization problems.The emergence of swarm intelligence algorithms solves this problem.It is simple,efficient,and effective.Adaptability and other advantages.Drosophila algorithm is a new swarm intelligence algorithm.The basic idea comes from the fruit fly foraging,the cooperative and competitive behavior between Drosophila groups,and uses its cooperative mechanism and information sharing mechanism to search for food locations.Since the introduction of the algorithm,it has attracted extensive attention from the academic and industrial circles and can effectively and quickly resolve related issues in practical engineering problems.This paper studies the operational flow,application fields and previous improvement strategies of the basic Drosophila algorithm,and proposes a Drosophila algorithm with self-adaptive reduction of the interval search step to optimize the SVM regression model.The improved Drosophila algorithm not only improves the convergence accuracy,but also increases the convergence speed and expands the scope of optimization.It effectively solves the problem of premature and easily falling into local extremes of the swarm intelligence algorithm.It can effectively solve the multiobjective,high-dimensional and complex optimization problems in the real industrial,scientific research and other fields.The research content and work of this article are as follows:First of all,carefully study the development status of Drosophila algorithm.In different fields,the improvement strategies made by previous generations in different directions,and describe the advantages and disadvantages of the algorithm flow and algorithm,and the influence of different parameter selection on the fruit fly optimization performance.Secondly,in-depth study of Drosophila algorithm,this paper proposes a Drosophila algorithm that adaptively reduces the search step size.The algorithm first expands the Drosophila population from a two-dimensional plane to a threedimensional space,increasing the search range and the size of the search space.Second,we changed the fixed search step size of Drosophila to a self-adaptively reduced search step length to achieve a dynamic balance between the previous global search ability and the late local search ability.Third,the determination of individual fruit flavour concentration determination values from the reciprocal of the individual to the origin distance is improved to relate to the reciprocal of the distance from the individual to the origin and the taste concentration determination value of the individual in the previous generation of the optimal Drosophila individual,so that the Drosophila individual search range is expanded.Wide range of real numbers greater than 1.Thirdly,this paper optimizes the parameters of SVM regression model for Drosophila algorithm with self-adaptively reduced step size.In-depth study of the mathematical theory of the support vector machine,the core function of the support vector machine-core function.The influence of related parameters and parameter values on the model in the support vector machine regression model is clarified.Finally,this paper takes the case of the corrosion status of the gas gathering pipeline in the Moxi Gas Field and the opening index of the Shanghai Stock Exchange Index.The Drosophila algorithm with self-adaptively reduced search step size was combined with the support vector machine regression model for applied research and verification.Compared with genetic algorithm,particle swarm algorithm and basic fruit flies algorithm,the results showed that the fly-flying algorithm with adaptively reduced search step size across generations has better performance in search optimization,faster convergence,and higher precision.Satisfactory results.