Fireworks Algorithm Based on Dual Population for Optimization Problems

A fireworks algorithm based on dual population is proposed to solve optimization problems. A dual population strategy is used to amend the shortcomings, i.e. slow convergence and bad population diversity of the existing fireworks algorithms. Each of two populations is running independently. Meanwhile, they alternately perform the hill-climbing and collaborative operator during the evolution process. Therein, the hill-climbing operator can enhance the local search performance of the proposed algorithm. And the collaborative operator is utilized to maintain the population diversity, avoiding getting stuck in local optimal regions. Furthermore, the proposed algorithm improves the setting of the maximum amplitude of the explosion and uses the tournament selection strategy to improve the convergence rate. The experimental results indicate that the proposed algorithm is superior to the compared algorithms in terms of the stabilization and reliability for most of test problems. It has higher accuracy lever and faster convergence rate.