Abstract
Most metaheuristic algorithms, including harmony search (HS), suffer from parameter selection. Many variants have been developed to cope with this problem and improve algorithm performance. In this paper, a hybrid algorithm of HS with grey wolf optimizer (GWO) has been developed to solve the problem of HS parameter selection. Then, a modified version of opposition-based learning technique has been applied to the hybrid algorithm to improve the HS exploration because HS easily gets trapped into local optima. Two HS parameters were automatically updated using GWO, namely, pitch adjustment rate and bandwidth. The proposed hybrid algorithm for global optimization problems is called GWO-HS. The GWO-HS was evaluated using 24 classical benchmark functions with 30 state-of-the-art benchmark functions from CEC2014. Then, the GWO-HS has been compared with recent HS variants and other well-known metaheuristic algorithms. The results show that the GWO-HS is superior over the old HS variants and other well-known metaheuristics in terms of accuracy and speed process.