A modified self-adaptive marine predators algorithm: framework and engineering applications

The application of metaheuristic algorithms is one of the most promising approaches for solving real-world problems. The marine predators algorithm (MPA) is a recently proposed population-based metaheuristic algorithm that has been proven to be competitive with other algorithms. Although the MPA shows good performance compared with other algorithms, modifications are still necessary to improve its optimization performance. Therefore, this paper proposes a modified MPA (MMPA). First, a logistic opposition-based learning (LOBL) mechanism is put forward to improve the population diversity and generate more accurate solutions. Second, effective self-adaptive updating methods are introduced into the original MPA, such as proposing the new position-updating rule, inertia weight coefficient and nonlinear step size control parameter strategy. The validity of the MMPA is tested on 23 classical benchmark functions, CEC 2020 functions and four real-world problems. Furthermore, the proposed algorithm is also evaluated using high-dimensional (Dim = 100, 1000 and 2000) benchmark functions. The experimental results and two different statistical tests demonstrate that the MMPA exhibits superior performance, and that it is competitive with many state-of-the-art algorithms in terms of accuracy, convergence speed, and stability.