Multi-step preferred elite-guided firefly algorithm
Xiaoyu Wei, Jun Li, Zhigao Zeng · 2021 IEEE 33rd International Conference on Tools with Artificial Intelligence (ICTAI) · 2021
Firefly algorithm has large position oscillation and high computational complexity. To address these problems, a multi-step preferred elite-guided firefly algorithm (MPEFA) is proposed. Firstly, the algorithm uses multi-step learning to select historical elite individuals. The elites will guide the movement of fireflies. Secondly, if the elite stagnates, the perturbation factor is used to optimize it to promote algorithm convergence. Finally, if a common firefly stagnates, the even-order Chebyshev mapping and opposition-based learning will be used to reinitialize the individual position to improve the ability of algorithm to jump out of the local optimum. Theoretical analysis proves the global convergence of the MPEFA. The simulation is carried out on the CEC2020 standard test function. The experimental results show that MPEFA is superior to other comparison algorithms in the unimodal, mixed and composite classes of numerical functions with a large amount of calculation, and has higher accuracy.