A novel metaheuristic algorithm with applications in parameter estimation and engineering problems

This paper develops a metaheuristic, the Natural Selection Optimization Algorithm (NSOA), taking inspiration from the natural selection and the survival, growth, and reproduction of the fittest. The algorithm uses two instruments: Normal distribution and Sigmoid function. The Normal distribution generates random numbers such that the individuals in the initial population obey a Normal distribution, which can place search agents appropriately. The Sigmoid function controls the search process of the NSOA, aiding the convergence of the algorithm. To appreciate the global search ability of the NSOA, we test the NSOA against 52 benchmark functions, a hydrogeologic parameter estimation problem, and three engineering problems. Moreover, this paper compares the NSOA with six other algorithms under the same experimental configuration. Our results show that, for most functions, the NSOA converges and extracts the optimal value faster than the other established algorithms, pointing to the novelty and practical efficacy of the NSOA.