The introduction of this strategy gives the algorithm good convergence and robustness, and enables it to find high-quality solutions stably under different problems and parameter settings.