A neighborhood information utilization fireworks algorithm and its application to traffic flow prediction

Abstract

A Fireworks Algorithm (FWA) is a novel intelligent optimization algorithm inspired by fireworks explosion, and it has been shown to be superior or competitive in various fields. Current FWA generates sparks without considering neighborhood information. This work improves sparks generation process by utilizing neighborhood information, and thus a neighborhood information utilization fireworks algorithm (NiFWA) is proposed. Specifically, fireworks are first divided into different subpopulations at different stages, and each subpopulation is considered as a whole, as well as it is also a neighborhood of other subpopulations. Then, a firework in a subpopulation generates its offspring by incorporating neighborhood information. Experimental results show that the proposed method is superior to FWA and its variants, competitive to state-of-the-art methods on CEC2013, CEC2017, three real world problems from CEC2010. Finally, NiFWA is used to assist kernel extreme learning machine in predicting the traffic flow.