Sine cosine algorithm with peer learning for global numerical optimization

Abstract

The sine cosine algorithm (SCA) is a simple and effective optimization method. However, it may exhibit stagnation behaviour in addressing complex optimization problems. Therefore, a novel improved version called the sine cosine algorithm with peer learning (PLSCA) is proposed in this study. According to the roulette wheel selection method, a peer learning strategy is designed to achieve information exchange between individuals. Then, a control parameter is introduced to control the combination of the peer learning strategy and original position updating equations, and this parameter can be adaptively adjusted by means of the individual evolution. To estimate the search capability of PLSCA, it is compared with six representative SCA methods and several other metaheuristic algorithms on two different kinds of benchmark function. Two chaotic time series problems are used to further evaluate the effectiveness of PLSCA. The results of simulation experiments reveal that PLSCA is a promising method.