The Improvement And Application Of Gravitational Search Algorithm

With the limitation of traditional numerical methods in practical applications,the swarm intelligence algorithm has made up for its shortcomings and become a research hotspot in recent years.Gravitational search algorithm,a new swarm intelligence algorithm,was proposed by Iran's scholar E.Rashedi and others in 2009 when they simulated the most common phenomenon of gravitation in nature.Gravitational search algorithm has the advantages of swarm intelligence algorithm,such as strong global search ability,relatively fast convergence speed and good practicability.At present,the algorithm has been successfully used to solve the problems of Web server selection based on QoS,linear and nonlinear filtering model,classification of multi classification data set.It also is efficient to solve other problems,such as estimation of decision function,fault diagnosis,identification of parameters in chaotic system,multi target economic decision,slope stability analysis and so on.But it has shortcomings.For example,the algorithm is prone to premature and the accuracy is not high.In this paper,gravitational search algorithm are improved,optimized and applied.The main work includes the following contents:(1)This paper proposes a gravitational search algorithm which dynamically adjusts the inertia weight and trend factors of speed and position(MACGSA).Adding dynamic inertia weight into the updating formula of mass makes weight have a nonlinear decreasing trend.So it improves the optimization precision and convergence speed of the algorithm.At the same time,the speed trend factor and the position adaptive factor are introduced.As the result of it,the algorithm can dynamically constrain the moving step of each generation of particles according to the number of iterations of the current population and make the algorithm adaptable.The mathematical analysis proves the convergence of the improved algorithm.The simulation results show that this algorithm has better optimization performance and higher convergence speed accuracy.(2)A gravitational search algorithm with adaptive mixed random mutation mechanism(MGSA)is proposed.The mutation triggering function with adaptive properties is introduced.This makes all particle individuals have the probability of mutation at any time.And the number of particle in the mutation in the population tends to decrease with the increase of the number of iterations.At the same time,the uniform mutation and Laplasse-Normal hybrid mutation are carried out in the whole optimization process of the algorithm.The uniform mutation makes the algorithm find the global optimal region quickly.And the hybrid mutation continues the depth search to improve the optimization performance.Theoretical analysis proves that the improved algorithm is convergent.Through simulation analysis,the improved algorithm achieves the better accuracy in solving function optimization problems,and the speed of optimization is also significantly improved.(3)The problem of unmanned navigation route planning transformed,a mathematical model of two-dimensional route planning and the reasonable evaluation performance are set up.The improved algorithm MGSA is applied to the route planning.Through the calculation examples and the comparison with other algorithms,the simulation results show that the MGSA algorithm has better evaluation performance and has obvious advantages in the convergence speed.So the effectiveness of the MGSA algorithm in solving the route planning problem is proved.