Explosion sparks generation using adaptive transfer function in firework algorithm

Fireworks algorithm (FWA) is a new Swarm Intelligence algorithm simulating the explosion process of fireworks. An improved Fireworks algorithm (IFWABS) with a transfer function is recently developed. In the IFWABS algorithm, a transfer function is used to map the sequence number in the sorted front of fireworks to a function value which is used to better calculate the number of explosion sparks as well as their amplitudes. This paper proposes an adaptive transfer function in which the differences in fitness values of fireworks are used instead of sequence number in the sorted front. The proposed method is applied on 28 CEC2013 benchmark optimization problems and a comparative study has been made with FWA, IFWABS and Enhanced Fireworks algorithm (EFWA). The experimental results show that the proposed algorithm statistically performs better than other algorithms.