Meta-heuristic algorithms have a great impact on modern optimization technology and engineering applications,and thus they have been widely used in different aspects of real-world applications.As a new meta-heuristic optimization algorithm,the flower pollination algorithm has the advantages of simplicity,high efficiency,flexibility and easy to implement.It provides an effective method for solving complex optimization problems and is successfully used to solve some engineering application problems.This paper proposed an improved algorithm of flower pollination and a discrete algorithm based on the basic flower pollination algorithm.The proposed algorithm is then applied to optimization problems.The main research results of this paper are as follows:(1)Introduce the biological background of the flower pollination algorithm,research significance and latest developments in this area.The basic theory of the algorithm,the algorithmic steps and the related parameter settings were presented.(2)A differential evolution flower pollination algorithm with a time-varying factor was proposed to improve the step-size factor,and the strategy of differential evolution was added in the iterative process to improve the convergence speed and optimization ability of the algorithm through population hybridization and mixing.Simulations have been conducted by using the standard test functions,and the results show that the TVDFPA is faster and more accurate.(3)The pressure vessel design problem is a typical constrained optimization problem.In this paper,the hybrid TVDFPA and double fitness value comparison method with a variable parameter setting have been used to solve the problem of pressure vessel designs.The experimental results show that the improved algorithm has better performance.(4)In order to solve the problem of combinatorial optimization,a Discrete Flower Pollination Algorithm(DFPA)was proposed.By redefining the concept of flowers,global search and local search,Lévy flight with appropriate segmentation.Experimental data and results show that the algorithm can find the optimal solution more quickly and accurately in solving the traveling salesman problem.The percentage of algorithm deviation is obviously reduced. |