Dynamic sine cosine algorithm for large-scale global optimization problems

Abstract

The sine cosine algorithm (SCA) is a recently proposed swarm intelligence optimization based on sine and cosine mathematical functions. It has a novel principle to process global optimization, but when solving large-scale global optimization problems, the performance of this algorithm is greatly reduced. To tackle this problem, a dynamic sine cosine algorithm (DSCA) is proposed. DSCA includes a nonlinear random convergence parameter to update equation, dynamically balancing the exploration and exploitation of SCA. Moreover, in order to avoid fall into the local optimum, a dynamic inertia weight strategy is introduced to modify the position equation of this algorithm. To evaluate the performance in solving large-scale global optimization problems, DSCA is compared with state-of-art algorithms. The 15 standard high-dimensional functions ranging from 200 to 5000 and IEEE CEC2010 functions are selected. The results show that DSCA has better convergence precision, faster convergence speed and stronger robustness when solving large-scale optimization problems. Two practical engineering problems are also applied to this algorithm, the effectiveness of the dynamic sine cosine algorithm to ensure the efficient results in real-world optimization problems.