Research On The Dragonfly Algorithm Based On Enhancing Individuals' Flight Direction And Its Application

Swarm intelligence optimization algorithm,originated from the study of artificial life,is an integral part in the field of artificial intelligence.It is gradually gaining prominence as more and more solutions to high complexity problems are required,which may be achievable within a reasonable period of time,but not optimal.Swarm intelligence algorithm solves complex optimization problems by simulating natural ecosystem mechanism.Due to the advantages of simple principle,few parameter adjustments and easy programming,swarm intelligent optimization algorithm is widely used in fields like combinatorial optimization,parameter estimation,function optimization,path planning,neural network training,graphics and image processing.The dragonfly algorithm is an effective swarm intelligence optimization algorithm,which is inspired by the behavior of the dragonflies' flight.It has been widely applied to solve single-objective and multi-objective problems,optimize support vector machine parameters,and feature selection problems.However,most swarm intelligence optimization algorithms have problems such as slow convergence speed,low algorithm accuracy,local optimal solutions,decline in population diversity in the later period of algorithm iterations and poor global searching capabilities.The dragonfly algorithm is no exception.In order to solve the weak global search ability,local optimal solution and low algorithm accuracy of the original dragonfly algorithm,this paper carries on a research,whose contents are as follows:(1)This paper proposes a dragonfly algorithm based on enhancing individuals' flight direction.On the one hand,the formula of inertia weight is changed.On the other hand,selection strategies are added when the location of individual dragonfly update.(2)The dragonfly algorithm based on enhanced individual flight direction is discretized to prepare for feature selection.(3)The proposed dragonfly algorithm based on enhancing individuals' flight direction is applied to function optimization and feature selection.In the experimental phase,this paper uses 13 benchmark test functions,and compares the results obtained by the proposed algorithm with the particle swarm optimization algorithm,gray wolf optimization algorithm,dragonfly algorithm,crow search algorithm and memory based hybrid dragonfly algorithm for verification.The statistical results demonstrate that the dragonfly algorithm based on enhancing individuals' flight direction outperforms other algorithms,in terms of high local optima avoidance ability and fast convergence speed.Furthermore,in order to demonstrate the applicability of the proposed algorithm in solving complex real-world problems,the dragonfly algorithm based on enhancing individuals' flight direction is employed to solve the feature selection problem as well.The dragonfly algorithm based on enhancing individuals' flight direction,as a feature selection approach,is tested on 18 data sets acquired from the UCI.The results reveal that the dragonfly algorithm based on enhancing individuals ' flight direction has comprehensive superiority in solving the feature selection problem,especially in the area of complementary medical diagnosis,which proves the capability of the proposed algorithm in solving real-world complex problems.