Multi-objective firefly algorithm based on compensation factor and elite learning

Aimed at early maturing and poor accuracy of multi-objective firefly algorithms, we propose a multi-objective firefly algorithm based on compensation factor and elite learning (CFMOFA). Based on iterations by introducing a compensation factor into the firefly learning formula, constraints by population can be overcome and the Pareto optimal solution can be approached in a reduced period. The non-inferior solutions produced in iterations were stored in the external archive and a random external archive particle was employed as the elite particle for population evolution. In this way, the detection range of firefly was extended and diversity and accuracy of non-inferior solution set were enhanced. The conventional algorithms, the improved algorithms and the proposed multi-objective optimization algorithm were tested and compared with each other. The results indicated great advantages of the proposed algorithm in convergence, diversity, and robustness and the proposed algorithm is an effective multi-objective optimization method.