A heuristic algorithm named the leader of dolphin herd algorithm (LDHA) is proposed in this paper to solve an optimization problem whose dimensionality is not high, with dolphins that imitate predatory behavior. LDHA is based on a leadership strategy. Using the leadership strategy as reference, we have designed the proposed algorithm by simulating the preying actions of dolphin herds. Several intelligent behaviors, such as “producing leaders,” “group gathering,” “information sharing,” and “rounding up prey,” are abstracted by LDHA. The proposed algorithm is tested on 15 typical complex function optimization problems. The testing results reveal that compared with the particle swarm optimization and the genetic algorithms, LDHA has relatively high optimization accuracy and capability for complex functions. Further, it is almost unaffected by the inimicality, multimodality, or dimensions of functions in the function optimization section, which implies better convergence. In addition, ultra-high-dimensional function optimization capabilities of this algorithm were tested using the IEEE CEC 2013 global optimization benchmark. Unfortunately, the proposed optimization algorithm has a limitation in that it is not suitable for ultra-high-dimensional functions.
View full abstract