Research On The Improvement Of Swarm Intelligence Optimization Algorithm

Optimization problems can be seen everywhere in our life and its core idea is to explore the optimal solution in the feasible range by designing appropriate schemes under specific conditions.With the development of technology,the constraints of problems are becoming more and more complex and the traditional methods cannot help.The birth of swarm intelligence algorithm for optimization just makes up for the above shortcomings.This kind of algorithm can better solve complex,nonlinear and large-scale problems and has the advantages of good flexibility,strong robustness and high efficiency of solution.Therefore,it has been researched and explored by scholars at home and abroad since it was proposed,and widely used in various fields.With the deepening of research,the types of swarm intelligence algorithms for optimization have been continuously expanded to solve new complex problems.Fruit fly optimization algorithm,flower pollination algorithm and whale optimization algorithm are the new swarm intelligence optimization algorithms proposed in recent 10 years.The three algorithms are simple in structure,easy to understand and implement.However,because it is short to bring up these the algorithms,and due to lack of mathematical theoretical basis,they have the disadvantages of slow convergence,easy to fall into local optima and low accuracy.In order to resolve these problems,scholars at home and abroad have improved the algorithms from search radius,optimization formula,parameter selection and other aspects,and finally improve the optimization performance of the algorithm and expand the application field.Nevertheless,with the increasing complexity of the optimization problem,the optimization ability of the algorithm needs to be constantly improved.Therefore,in order to further improve the optimization performance of the algorithm,this paper proposes three new improved algorithms based on the previous research,and the improvement measures are as follows:(1)Fruit fly optimization algorithm with dynamic adjustment of search strategy(FOAASS): Firstly,the initial position distribution is improved by chaotic mapping so as to improve the quality of the initial solution;Secondly,the convergence rate is improved by predicting the evolution direction of the population;Thirdly,the search ability is enhanced by randomly selecting the search radius with dynamic adjustment;Finally,the search strategy is dynamically adjusted to jump out of the local best.(2)Flower pollination algorithm based on dynamic adjustment and cooperative search(FPADC): Firstly,the initial position distribution is improved by Holden sequence to improve the quality of the initial solution;Secondly,the convergence rate is increased by refining the individual division of labor within the population;Finally,the optimization accuracy is improved by balancing the global search and local exploration capability with dynamically adjusting the conversion probability and optimization formula.(3)Dynamic search and cooperative learning for whale optimization algorithm(DCWOA): Firstly,the quality of the initial solution is improved by equivalent substitution and Faure sequence;Secondly,the evolutionary direction of the population is guided by elite individuals and the local optima are jumped out by the mutation strategy;Finally,the algorithm search capability is improved through dynamically adjusting the convergence factor and search equations.In order to better verify the effectiveness of the above improvement measures,this paper selects a number of test functions to carry out simulation experiments on the algorithms in various dimensions: Firstly,through the fixed number of iterations of experiments,the optimization accuracy and convergence speed of various algorithms are analyzed and compared;Secondly,through the fixed convergence accuracy of experiments,the average number of iterations and experimental success rate of various algorithms are analyzed and compared.The experimental results show that the three improved algorithms proposed in this paper have better optimization performance compared with the improved algorithms in the references.