Recent Developments of Fireworks Algorithms

Recent Developments of Fireworks Algorithms

DOI: 10.4018/978-1-7998-1659-1.ch001
OnDemand:
(Individual Chapters)
Available
$37.50
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Fireworks Algorithm has been proposed for almost 10 years. Because of its basic but profound collaborative searching manner and advantages of universal effectiveness, hundreds of scholars have conducted and published a wide range of work on it. This chapter serves as a background description of fireworks algorithms' developments by introducing its detailed status of researches and applications. Specifically, it gives a brief summarization and analysis of published researches on Fireworks Algorithms since 2010 to clarify characteristics of its historical progress and future trend in detailed fields like algorithms improvements, theoretical analysis, and practical applications.
Chapter Preview
Top

Fireworks Algorithms

Motivation and Principle

Fireworks and firecrackers are one of the traditional events for Chinese festivals, especially for New Year’s Eve. In the night, fireworks rise and explode, bursting with plenty of sparks to light up the sky. Plenty of fireworks explode in different ways which form a distribution of sparks over the night sky. Such a process has much in common with the optimization process. And this is how fireworks algorithm been inspired and proposed at 2010 by Prof. Tan and Zhu (Tan & Zhu, 2010).

Figure 1.

Fireworks in the night sky

978-1-7998-1659-1.ch001.f01

Most meta-heuristic algorithms designed delicate local optimization mechanisms to accelerate population convergence. However, the behavior of fireworks illustrates a different perspective for global optimization, that is, to manage several simple local search process conducted by sub-groups of individuals and enhance global optimization by collaboration between each group.

So the population of fireworks algorithms are composed of several individual called fireworks and each firework corresponds to plenty of individuals called sparks. In each iteration of the optimization, each firework simply generates certain number of sparks around itself. However, the distribution and allocation of sparks are carefully designed with considering of cooperation between fireworks. With proper collaboration method, fireworks algorithms are able to achieve stable global exploration and fast local exploitation with very basic search method for each firework. The framework of fireworks algorithms can also be viewed as cooperation of several simultaneous local optimization process, thus could be easily combined with other methods.

Practically, fireworks algorithms simply scatter sparks with uniform distribution or normal distribution around itself. For fireworks with better fitness (objective value), more sparks are usually generated within closer range. For fireworks with worse fitness, less sparks are usually generated within farther range. The process of generating sparks for each firework is called explosion. And the next generation of fireworks are selected from current fireworks and sparks.

Complete Chapter List

Search this Book:
Reset