Adaptive Parallel Flower Pollination Algorithm

Basic flower pollination algorithm (FPA) has some disadvantages, such as weak local search ability, slow convergence speed and low convergence accuracy. This paper proposes an improved adaptive parallel flower pollination algorithm. First, the parallel mechanism is introduced into the FPA to improve the shortcomings of insufficient diversity of a single population in the middle and late stages of the calculation. As a result, the parallel mechanism with different parameter values can further strengthen the algorithm’s global search ability while keeping the same local search ability. Second, a nonlinear algorithm behavior conversion probability P is used, and a nonlinear Levi flight step scale factor is added. The former enables the algorithm to dynamically control global and local pollination behaviors according to the calculation period; the latter enables the algorithm to respond to the processing status and adaptively adjusts the jumping step length of pollen individuals in the solution space. In all, the two mechanisms together strengthen the search ability of the algorithm and the ability to jump out of the local optimum. The analysis of the optimization results of the test functions shows that compared with some other algorithms, the proposed algorithm greatly improved the accuracy of the optimal solution and the convergence speed of the optimization.