IMPROVED BAT ALGORITHM WITH NOVEL SEARCH MECHANISM AND ONE-DIMENSIONAL PERTURBATION LOCAL SEARCH STRATEGY

摘要

Bat algorithm often suffers from premature convergence, poor convergencerate and low accuracy in solving high-dimensional optimization problems because of quicklylosing diversity. To enhance population diversity and sea/rch accv/racy and speed upconvergence rate of bat algorithm, this paper proposed a novel search mechanism andone-dimensional perturbation strategy to balance the ability of exploration and exploitationduring the search process. Novel pulse emission rate, loudness, velocity and locationupdating functions are designed to avoid premature convergence. A novel population movingmechanism is introduced into bat algorithm to help the trapped bats escape from localoptimum and novel one-dimensional perturbation local search strategy is designed to increaseefficiency and accuracy of local search. The proposed novel improved bat algorithmhas been evaluated on a set of typical nonlinear benchmark functions and compared withrecently improved bat algorithms and other evolutionary algorithms. Experimental resultsand statistic analysis confirm promising performance of the improved bat algorithm insolving high-dimensional nonlinear functions.