Research Of Improved Firework Algorithm And Application

In 2010,Professor Tan Ying published "Fireworks algorithm for optimization" at the first International Swarm Intelligence Conference,and proposed the groundbreaking theory of fireworks algorithm.The firework algorithm has a strong local search ability and a global search ability by self-adjustment mechanism,which can deal with the solution of very complex optimization problems.Through direct or indirect collaboration and interaction between individual fireworks,the overall complexity is shown.Intelligent behavior,which has developed a large number of global optimization methods,has become a research hotspot in recent years.The number of sparks generated by the explosion of the firework algorithm is unstable under different objective functions is the defect of Firework Algorithm.The lack of information exchange mechanism between particles,and the large randomness of the next generation of selection are also defects of Firework Algorithm.Aiming at the problem this paper proposes improvements to the explosion operator,mutation operator and selection operator.Furthermore,an Explosive communicative selected firework algorithm(ECSFWA)is proposed.The algorithm determines the number of explosion sparks by adopting a better fitness value sorting level,adding a trending movement to the optimal firework particles of the current population to the firework particles with a certain probability,adopting an adaptive scale to select the next generation of fireworks,and using the elite strategy retains the optimal firework particles,which makes the improved ECSFWA more reasonable and more efficient.The steps and process of the ECSFWA algorithm are given,and the convergence of the algorithm is proved.Then using several sets of classic test functions to compare with other classic improved firework algorithms,the effectiveness of the ECSFWA algorithm is verified.Finally,the ECSFWA algorithm is applied to the solution of the clustering problem,and the distance within the class and the fitness function are set.Furthermore,the ECSFWA algorithm,the DYNFWA algorithm,and the EFWA algorithm are used to cluster the Iris data set and the Wine data set to verify the algorithm's performance,the practical application ability of the algorithm is verified.