An enhanced ivy algorithm fusing multiple strategies for global optimization problems

Increasingly sophisticated science and technology are always accompanied by the emergence of optimization problems of higher complexity. To provide a new and higher performance optimization technique, this paper proposes a novel enhanced ivy algorithm (AFDIVYA, for short) that mixes multiple strategies. In AFDIVYA, two strategies are specifically designed for IVY (the adaptive perturbation factor and the adaptive growth rate). The adaptive perturbation factor enhances the ability of the population to explore locally. And the adaptive growth rate contributes to the balance between the exploration and exploitation ability. In addition, the fish population device strategy and the differential evolution strategy are introduced for the first time. The two strategies effectively enhance the diversity of population and expand the search space. To verify whether the fusion of the four strategies enhances the accuracy of IVYA in solving problems, this paper sets up multiple sets of experiments. First, for relatively high dimensional problems, AFDIVYA is compared with several excellent meta-heuristic algorithms on 30, 50, and 100 dimensions of CEC2020. AFDIVYA performs the best with an average ranking of 2.2, 2.1, and 1.6, respectively. For low-dimensional problems, the same comparison algorithms are tested on CEC2022. AFDIVYA also has the smallest average ranking of 2.4. What's more, Wilcoxon rank sum test proves the validity of the AFDIVYA proposal. And then this paper selects several complex engineering applications of different dimensions to test the ability of AFDIVYA to cope with real complex problems with constraints. The results demonstrate the excellent precision and accuracy on these complex problems. Finally, a more challenging and more relevant shape optimization problem is applied. AFDIVYA is tested and unsurprisingly has an excellent performance. In conclusion, AFDIVYA is very competitive among many existing metaheuristics. And this paper provides an advanced technique for solving more challenging real-world problems in the future.