Triangle Search Optimization Algorithm for Single-Objective Bound-Constrained Numerical Optimization

Real-parameter optimization has been a focus of the last decade. An inspired algorithm, triangle search optimization (TSO), is proposed for the Congress on Evolutionary Computation (CEC) 2020 competition. In this paper, the TSO algorithm is divided into two phases: the triangle vertex searching (TVS) and triangle edge searching (TES) phases. In the TVS phase, the population is divided into two subpopulations, which can be enhanced by vertex searching operators and the covariance matrix adaptation evolution strategy (CMA-ES) for exploitation. In the TES phase, the differential evolution vector between superior and inferior solutions is employed to improve the exploration ability. The TSO algorithm is evaluated on the CEC 2020 benchmark problems and compared with the winners of the CEC competition and novel competitors. The experimental results show that TSO outperforms the compared algorithms.