An intensify atom search optimization for engineering design problems

Atom search optimization is a newly developed metaheuristic algorithm inspired by molecular basis dynamics. The main changes of premature convergence and poor balance between exploration and exploitation still persist, which cannot yet do well in solving some complex optimization problems. To solve the above problems and get better performance, an improved atom search optimization with nonlinear inertia weight factor, neighbor learning component and updating mechanisms is proposed in this paper. First, to balance exploration and exploitation better, a nonlinear inertia weight factor is proposed to determine how much the previous velocity and acceleration is preserved. Second, to effectively increase the exchange of information among atoms, a neighbor learning component is integrated into the velocity updating equation, which is beneficial to enhance the exploitation ability. Thirdly, to help atoms jump out from the local optima, an updating mechanism of the best and worst atoms of the current population is proposed, and a greedy selection strategy is employed to boost the probability of finding the optimal solution. Finally, the well-known CEC2017 benchmark functions and four real-world engineering design problems were employed to demonstrate the performance of our algorithm. The experimental results and statistical analysis show that the performance of our algorithm is better than the comparison algorithms. (C) 2020 Elsevier Inc. All rights reserved.