Fireworks algorithm based on dynamic search and tournament selection

Abstract

The traditional fireworks algorithm (FWA) is easy to fall into the local extremum and the accuracy of the solution is not high as that the explosion amplitudes and the selection mechanism are inefficient. A FWA based on dynamic search and tournament selection is proposed. The algorithm redefines the original explosion amplitudes, introducing the concept of minimum explosion amplitudes. In order to make the explosion amplitudes calculated dynamically, the number of iterations and the maximum number of iterations are introduced into the updating formula. Meanwhile, without changing the physical meaning of the original algorithm, the formula preserves the fitness value. Updating the new explosion amplitudes by non-linear method, thereby, it achieves the goal of the global search in the early phases and the local search in the late part phases. At the same time, the tournament selection mechanism is used to select some excellent individuals from the population. And then it selects the best one to enter the offspring population until the new population size reaches the population size. It makes the result closer to the desired results and improves the accuracy of the solution. The proposed algorithm balances the global and local search abilities, making the calculation results more accurate.