A Novel Sine Cosine Algorithm for Global Optimization

Abstract

Sine cosine algorithm (SCA) has a fast convergence speed and is easy to implement. In order to overcome the evolutionary stagnation of swarm, this paper presents a novel SCA (NSCA) in which three learning strategies are used to update individuals and a selection mechanism is developed to guide each individual to choose a proper updating strategy. The selection mechanism is designed by the credit assignment method and Upper Confidence Bound (UCB). The proposed algorithm has been experimentally validated on 18 benchmark functions. Compared with SCA variants and other swarm intelligence algorithms, experimental results show NSCA is competitive in solving most functions.