In order to enhance the global optimization ability of bat algorithm

a new global"optimizing strategy"is proposed

which introduces the individual history optimum of bats into the optimizing strategy of individual bats and introduce a"guide"direction to guide the global optimizing of individual bats. Bat individuals will implement the strategy of"survival of the fittest"after each iteration

randomly initialize some bat individuals to replace individuals with lower fitness values

and increase their diversity to prevent them from falling into the local extreme value. The Wine

Sonar

Glass datasets in UCI are selected for clustering test. The experimental results show that compared with the original bat algorithm

differential evolution algorithm and particle swarm algorithm

the improved bat algorithm can get a higher clustering accuracy and stronger robustness.