Improved Fruit Fly Optimization Algorithm With Changing Step And Strategy And Its Application

The Fruit Fly drosohila Optimization Algorithm Optimization(FOA)is a new evolutionary algorithm for global optimization based on the evolution of fruit flies'foraging behavior.First,the corresponding taste concentration is calculated by the location of the fruit fly.Subsequently,the taste concentration decision is used as the independent variable of the fitness function to obtain the taste concentration value of each fruit fly.Take the maximum concentration of the taste as the current optimal value,and find the optimal value in the continuous iteration until the optimal value converges or runs out of all iterations.The FOA algorithm has the advantages of good stability,simple process and fast convergence.But the FOA algorithm has some disadvantages.First,the distance and direction of the movement of the fruit fly during the foraging process is random,and stride length is the only key factor.Because the step size is fixed,the search ability of the algorithm is limited to some extent,and the global performance and local performance can not be well balanced.Secondly,in many multimodal and multidimensional problems,the FOA algorithm is often trapped into local optima,which affects the overall performance of the algorithm.In view of these shortcomings of FOA algorithm,this paper makes some improvements:1.During the iterative process,two historical best values are selected randomly,and the difference plus a constant is the step size of the contemporary fruit fly population,which balances the global and local search capabilities of the whole population.When the Drosophila population becomes stable,a certain number of individuals of the fruit fly are mutated,using the median of the optimal value,the second and the third values on each dimension as the initial position of its variation.It can effectively avoid the disadvantage that FOA is easy to fall into local optimum,and improve the convergence speed and precision of the algorithm.2.The simulation results of fourteen common benchmark functions show that the convergence speed,the optimization accuracy and the stability of the proposed algorithm have been greatly improved.3.In order to further apply the improved algorithm to practical problems,the paper applies the algorithm to two very classic NP-hard problems-the 0-1 knapsack problem and the traveling salesman problem(TSP).First,8 classic 0-1 knapsack problems are used to test the performance of the improved algorithm.Since the dimensions of the 8 0-1 knapsack problems are between 10 and 100,the performance of the new algorithm in the 0-1 knapsack problem can be tested comprehensively.Secondly,we applied six data sets to the application of the new algorithm in the traveling salesman problem,the six datasets are very different in number of cities.The paper compares the improved algorithm with the particle swarm algorithm,and theoretically analyzes some of the performance of the proposed algorithm.Finally,it summarizes the research and application of this paper,and illustrates the advantages and disadvantages of the whole work.And according to the research and improvement examples of the optimization algorithms of fruit flies in recent years,this paper sorted out four points about the FOA algorithm which is worthy of further research.