Research On Firefly Algorithm And Its Application

As a typical swarm intelligence optimization algorithm,Firefly Algorithm(FA)has a wide range of applications in the fields of computer,engineering,management,and economics because of its good search ability.FA is a stochastic optimization algorithm designed to solve the optimization problem by simulating the biological characteristics of the firefly population.It has the advantages of simple implementation,few adjustment parameters,and strong practicability,but also has the disadvantages of slow convergence,easy fall into local optimization,and weak discrete optimization.Therefore,the algorithm still has much room for improvement.The main work of FA improvement in this paper is as follows:(1)Aiming at the step size and attractive part of the attracting movement formula of the firefly optimization algorithm,an adaptive step size and an improved strategy of proportional model attraction are proposed.In the adaptive step strategy,a control body that controls the change of the step size is added to determine whether the next step size is changed based on the current optimal individual position distance from the previous generation.When the distance is greater than a set threshold,the step size does not change.On the contrary,the step size is reduced according to the step size change formula to meet the demand for the step size in the current stage.In the improvement of the attractiveness of the scale model,due to the rapid convergence of the attractive part of the original standard firefly optimization algorithm,the diversity of population evolution in the middle and late stages is greatly reduced,and the improved FA will reduce the attractiveness of the current generation.It was changed to 0.4 times of the previous generation,and the population diversity was preserved as much as possible while meeting the needs of evolution.Finally,by comparing the experimental results of the improved FA optimization standard test function,it is shown that the above two improvement strategies do help improve the performance of FA.(2)Firefly algorithm based on opposition-based learning.Although the improved attractive part helps to preserve the diversity of the population,due to the demand for the accuracy of the solution results,the algorithm will inevitably reduce the diversity of the population during the evolution process and fall into a local optimum.Therefore,it is necessary to introduce a mechanism to increase diversity to improve the global search capability of the algorithm.To this end,the opposition-based learning strategy is introduced in Chapter 4 of this paper.The three individuals with the worst fitness values in each generation of the population are implemented opposition-based learning strategy,so that the worst three individuals use the central symmetry point of the solution space as the reverse point to jump out of the local optimal,and improve the diversity of the population.Experiments show that FA has improved its global search ability after introducing opposition-based learning strategy.(3)Because the firefly algorithm was mainly proposed for continuous optimization problems in the early stage,the ability to solve discrete problems is weak.In order to make it applicable to the discrete problem of flexible job shop scheduling,an individual discrete coding solution for fireflies is proposed.By transforming the processing sequence information of different workpiece processes into the position information of individual fireflies,the discreteness of the individual populations is resolved,and the target optimized time performance is given an adaptive value to the individual fireflies.After the result of the algorithm is obtained,the individual position information is decoded into the processing information of the machine,and the processing machine and start time of the workpiece process are determined.Finally,simulation tests verify the effectiveness of the improved coding design algorithm for solving flexible job shop scheduling problems.