The flower pollination algorithm is a heuristic algorithm to simulate the process of flowering plants.This algorithm uses plant's pollination process and self-pollination process to draw an analogy as global search and local search.It is used to solve various optimization problems because the algorithm has the advantages of simple realization,strong robustness and fast optimization.However,with the increase of the size and dimension of the processing data,the flower pollination algorithm has a slow convergence rate in the later period and easy to fall into the local optimal,followed by a problem of low optimization accuracy.To solve the above problems,a flower pollination algorithm based on improved harmony search and refraction principle is proposed in this paper.The algorithm firstly introduces the cross mutation operation in differential evolution into the harmonic search algorithm.The algorithm improves the disadvantage of strong randomness and slow speed in the optimization of the step length by dynamically adjusting the scaling factor in the differential evolution algorithm.Then,the harmonic search algorithm based on differential evolution is introduced into the flower pollination algorithm.The algorithm is used to find the current optimal solution.As the initial solution of the flower pollination algorithm,the quality of the initial solution and the convergence speed of the algorithm can be improved.Finally,the principle of refraction is introduced into the flower pollination algorithm,and the refraction point is obtained by refraction of the optimal solution of the current iteration.This process can expand the search range of the algorithm,enhance the diversity of the population,help the algorithm to get rid of the local optimal and improve the optimization and convergence accuracy of the algorithm.Simulation results show that the proposed algorithm is much better than the standard flower pollination algorithm. |