Differential evolution algorithm has been widely used, because of its efficient optimization and no complex operation and coding mechanism. But DE falls into the local optimum easily. So this study puts forward a memetic algorithm. The algorithm can increase the diversity of population and jump out the local extreme value point effectively. The convergence speed of the algorithm is improved, because of the selfadaptive operator which can adjust the scaling factor adaptively. Three classic benchmarks functions are used to evaluate the MMADE and basic differential evolution algorithm. The experimental results show that the MMADE algorithm is effective.