The Research Of Improvement And Application Of Artificial Fish Swarm Algorithm Based On Local Search

The Artificial Fish Swarm Algorithm utilizes a new bottom-up optimization model and adopts several typical behaviors of fish swarm: foraging,clustering and following.It has the characteristics of parallelism,simplicity and fast optimization,also it can effectively solve unconstrained continuous single-objective optimization problems.Although the time for the AFSA is not long,it has attracted the attention of many experts and scholars.At present,the AFSA has been improved in many aspects and applied in many fields,the research of it still has a good prospect.In AFSA,the use of random factors does not guarantee that the artificial fish is better than the pre-update position after each iteration.However,both the steepest descent method and the conjugate gradient method can ensure that each iteration is in the descending direction.Therefore,the artificial fish that has not been improved in the AFSA is updated by using the steepest descent method or conjugate gradient method,an improved artificial fish swarm algorithm based on gradient information is obtained.This ensures that all artificial fish positions after each iteration are better than they were before.In the AFSA,each artificial fish will affect other artificial fish within the perceptual range through the clustering operator and the following operator.Use the steepest descent method to update the best artificial fish in the school of fish,so that the best artificial fish position is closer to the optimal solution.Through the best artificial fish to transmit its own information,other artificial fish within its perceptual range are moved to better positions.Then this part of the artificial fish affect more artificial fish until the entire fish population,the elite accelerated artificial fish swarm algorithm is obtained.Combining the first two ideas of improvement,the conjugate gradient method is used to update the artificial fish that are not improved in the algorithm and the best artificial fish in the fish population.This not only ensure that all artificial fish move in a better direction,but also enhance the best artificial fish's ability to guide fish.A modified artificial fish swarm algorithm using conjugate gradient method is acquired.The selection of visual field and step size has great influence on AFSA.When the visual field is small,the artificial fish has stronger search ability in the adjacent area,and when the visual field is large,the artificial fish can easily find the global extreme value.When the step size of artificial fish is large,it is advantageous to converge to the extreme point.On the contrary,when the step size is small,the accuracy of the optimal solution obtained by the algorithm will be improved.With the development of the artificial fish swarm algorithm,reduce the field of vision and step size according to the number of iterations,so that the artificial fish's local search ability is strengthened,and the appearance of the oscillation phenomenon is reduced,an artificial fish swarm algorithm with improved step size and visual field is obtained(Algorithm LEAFSA).The numerical experiments of five improved artificial fish swarm algorithms were conducted using the typical unconstrained optimization test problem.The results show that the accuracy of the five improved artificial fish swarm algorithms compared with the basic artificial fishswarm algorithm and some other improved artificial fish swarm algorithms is improved,and the computation of the algorithm is less.We select the traditional Dengue virus transmission model and the fractional nonlinear dengue virus propagation model under the definition of Caputo fractional derivative,optimize the parameters of the models by using the Algorithm LEAFSA.The numerical results show that the parameters obtained by the Algorithm LEAFSA make the root mean square error between the output of the corresponding model and the actual data are very small.