An Improved Mayfly Optimization Algorithm Based on Quasi-Oppositional Learning

Random guesses are usually adopted in mayfly optimization algorithm (MOA) when prior knowledge about the solution is absent. The distances between the guesses and the optimal solution have a great impact on the convergence speed and solution accuracy. In this paper, an improved MOA based on quasi-oppositional based learning (QOBL) is proposed, called it as QOBLMOA. In the proposed QOBL-MOA, QOBL is introduced into the population initialization and position update process of MOA. In the initialization stage, the quasi-oppositional position of each mayfly’s position is generated. The best one between the initial position and its corresponding quasi-oppositional position is selected as the final initial position of the mayfly. Similarly, in the position updating stage, a quasi-oppositional position of each mayfly’s current position is generated, and the best one between them is selected as the final current position of the mayfly. Since the QOBL has the potential to explore a position nearer to the optimal solution than random guess, the proposed QOBL-MOA not only has faster convergence speed but also has a larger probability to jump out from the local optimum. The proposed QOBL-MOA is evaluated on 16 benchmark functions and 4 engineering design problems. Experimental results confirm QOBL-MOA performs better than other meta-heuristic algorithms.