关键词: 花授粉算法, 惯性权重, 非均匀变异, 收敛能力, 多样性

Abstract: In order to solve the problems of low accuracy computation, slow velocity convergence and premature convergence caused by the shortcomings of search equations of the flower pollination algorithm, it proposes a Flower Pollination Algorithm with New pollination Methods（NMFPA）. The global pollination in the algorithm is composed of the inertia weight and two groups of random individual difference vectors, to maintain the diversity of the population and improve the global exploration ability of the algorithm. A new local pollination strategy is constructed by using the information sharing mechanism and two new mutation strategies to enhance the local development ability of the algorithm, the optimal individuals based on Gauss mutation are used to guide the evolution direction of the other populations which reduces the blindness of individual evolution and improves the convergence speed and precision of the algorithm, and non-uniform mutation mechanism is introduced to increase the diversity of the population that avoids the algorithm is easy to fall into local extreme points and enhances the overall performance of the algorithm. The numerical simulation experiments are carried out on 22 test functions. The experimental results and statistical analysis show that the new algorithm has higher convergence precision and speed compared with the standard FPA, and can solve the premature problem better. In addition, compared with the improved FPA algorithm from many perspectives, the experimental results show that the improved algorithm proposed in this paper is a new and competitive algorithm. Meanwhile, the NMFPA algorithm is used to solve the permutation flow shop scheduling problem. The experimental results show that the new algorithm is feasible and has certain advantages for solving practical engineering problems.

Key words: Flower Pollination Algorithm（FPA）, inertia weight, non-uniform mutation, convergence ability, population diversity