A hybrid multiverse optimisation algorithm based on differential evolution and adaptive mutation
Lei Chen, Lvjie Li, Wenyue Kuang · Journal of Experimental & Theoretical Artificial Intelligence · 2020
Multiverse optimisation (MVO) algorithm is an excellent meta-heuristic algorithm based on laws of physics. However, it is easy to fall into local optimum when solving complex multimodal optimisation problems with high dimensions. The global optimisation performance of the algorithm is still unsatisfactory. In this paper, we propose a hybrid multiverse optimisation (DE-SMVO) algorithm. First, in order to increase the ability of information exchange and global exploration, a differential mutation strategy is introduced to improve the search equations of MVO based on the differential evolution (DE) theory, enabling the algorithm to explore more unknown space. In addition, the adaptive mutation is carried out on the current global optimal universe in each iteration process, and excellent mutation universe is retained, which enhance the exploitation ability and improve the convergence accuracy of MVO algorithm. In order to investigate the performance of the proposed DE-SMVO algorithm, it is evaluated on 23 benchmark functions in this paper. The experimental results prove that the optimisation performance is significantly better than that of the MVO algorithm, and its performance is also superior compared with nine state-of-the-art meta-heuristic optimisation algorithms.