Abstract: We presented a fast differential evolution (FDE) algorithm. The algorithm used the technique that constantly updated and narrowed the search area to determine the search interval of the next generation according to the previous generation optimal individual, so as to speed up the convergence rate and to improve the convergence precision and robustness. Through simulation test analysis of 21 extreme functions about optimization, the results show that the convergence rate, convergence robustness, and convergence precision of the algorithm are significantly superior to the other algorithms for the high dimensions of the problem, and the initialization form of population does not have any effects on the convergence performance of the algorithm.

Key words: algorithm； convergence precision； robustness； convergence rate, fast differential evolution (FDE)