An improved grasshopper optimization algorithm(GOA)combining improved DBSCAN clustering and multiple evolutionary strategies was proposed to overcome the shortcomings in complex high-dimensional problems,such as low convergence accuracy,weak optimization ability,and difficulty in jumping out of local optima.Firstly,by introducing multi-core weighted distance measurement and dynamic parallel operation strategy,the clustering efficiency of DBSCAN(density based spatial cluste-ring of application with noise)for high-dimensional data was improved.Secondly,utilizing the advantage for arbitrarily shaped datasets,the DBSCAN was used to analyze the GOA population,and the grasshoppers were endowed with spatial attributes such as core points,boundary points,and isolated points.Finally,taking into account the spatial characteristics of the population and the differences in the degree of evolution between individuals,various individual evolution strategies for locusts were designed to improve the global optimization ability of GOA.The simulation results of typical complex,high-dimensional test functions and classic TSP problems show that the improved GOA has more advantages in convergence accuracy.
NETL
NSTL
万方数据
