Enhancing three variants of harmony search algorithm for continuous optimization problems

Meta-heuristic algorithms are well-known optimization methods, for solving real-world optimization problems. Harmony search (HS) is a recognized meta-heuristic algorithm with an efficient exploration process. But the HS has a slow convergence rate, which causes the algorithm to have a weak exploitation process in finding the global optima. Different variants of HS introduced in the literature to enhance the algorithm and fix its problems, but in most cases, the algorithm still has a slow convergence rate. Meanwhile, opposition-based learning (OBL), is an effective technique used to improve the performance of different optimization algorithms, including HS. In this work, we adopted a new improved version of OBL, to improve three variants of Harmony Search, by increasing the convergence rate speed of these variants and improving overall performance. The new OBL version named improved opposition-based learning (IOBL), and it is different from the original OBL by adopting randomness to increase the solution's diversity. To evaluate the hybrid algorithms, we run it on benchmark functions to compare the obtained results with its original versions. The obtained results show that the new hybrid algorithms more efficient compared to the original versions of HS. A convergence rate graph is also used to show the overall performance of the new algorithms