The Improvement On Differential Shuffled Frog Leaping Algorithm And Its Application

Computer intelligent optimization algorithm imitates the natural wisdom and human wisdom,and it obtains a lot of attention from many researchers because of its intelligence,parallelism and robustness,and its good adaptive ability and strong global searching capacity.SFLA is a new intelligent optimization algorithm,which combines the advantages of local heuristic search of memetic algorithm and global search of particle swarm optimization algorithm.First,in the process of evolution,performing the exact local search,then conducting global search taking advantage of the information shared among the subgroups,by which they are combined with each other until find the global optimal solution.The structure of SFLA is simple and easy to understand,the control parameters are few and SFLA has a powerful global search ability.Differential evolution algorithm is also a new global optimization algorithm,and the local update strategy is similar to genetic algorithm,and it use differential mutation operation,crossover operation and selection operation update generate new individuals.After repeatedly local evolution,the search direction of the algorithm is close to the global optimal solution.DE algorithm has a exact local search ability and a strong robustness,and it has become a significant branch of intelligent optimization algorithm.Presently,combining local update strategy of differential evolution algorithm and other optimization technique has been widely applied in various fields,and it play an important role in scientific research and practice.This article is focused on the disadvantages that SFLA is easy to fall into local optimum and premature convergence in the optimization process,and presents a DSFLA which combines the exact local search ability of differential evolution algorithm and the powerful global search ability of leapfrog algorithm.This algorithm draws lessons from the thought of variation in differential evolution,in earlier period,it use the useful information of other individuals in subgroup to update the worse individual and increase local disturbance,which will increase the diversity of population,and in later period,use of the best individual information for mutation crossover operation to speed up the convergence.At the same time,after each generation of new individual,it needs a improved cross-border processing to dynamically adjust the scale of change,and then selects the most suitable individual compared with the selection of the worst individual among the subgroup.In this article,the diversity of the population is further preserved by using the archive set.The results of the simulation of five typical continuous optimization functions showed that DSFLA is better than SFLA and SFLA-AV not only in the stability of getting the optimal solution,but also in quality.It played a very good effect in maintaining the diversity of population in earlier period and speeding up the convergence to avoid premature algorithm in later period.Finally,in this article,the DSFLA is used to solve vehicle routing problem with capacity constraints in logistics.The real coding is used to initialize the population,and the rule of DEB is used to solve constraint problem,and by experimental simulation varied optimal paths are concluded,which can provide various scheduling schemes for practical logistics problems.