Orthogonal learning harmonizing mutation-based fruit fly-inspired optimizers

The original fruit fly optimizer (FOA) has two core disadvantages: slow convergence speed and low solution quality. Furthermore, fruit fly optimizer tends to skip the optimal optimum when faced with complex or high-dimensional problems. To overcome these shortcomings, we introduce Gaussian mutation and orthogonal learning schemes into the fruit fly optimizer. On the one side, the orthogonal learning strategies can acquire more useful information during the exploratory and exploitative stages and build superior lead vectors. On the other hand, the Gaussian mutation mechanism also increases the population's perturbation and enhances the diversity of the swarm. With these mechanisms, the proposed method has a higher potential to avoid premature convergence and fall into local optimum. To validate the performance of the proposed method, it is compared with three other state-of-the-art variants of fruit fly optimizer over several reprentative benchmark functions. The results have demonstrated the efficacy of the proposed method is superior to the conventional fruit fly optimizer according to both convergence rapidity and solution quality. Simulations reveal that the proposed new FOA variant has more stable performance and high potential. (C) 2020 Elsevier Inc. All rights reserved.