Aimed at the problem that Baldwin effect in memetic differential evolution is not ripe for application, This paper proposes a Baldwin effect-based memetic differential evolution (BMDE) algorithm. The algorithm takes the simplified Hooke Jeeves as a local search and DE for globe search with Baldwin effect to Differ from other memetic DE algorithms. The proposed algorithm uses a new method to carry out Baldwin effect by enlarging learned probability of individuals with better local search to change the evolution direction and diversify the population. Tested by 30 benchmark functions in CEC2014 and compared with standard DE and 3 state-of-the-art DE algorithms, BMDE performs satisfied convergence ability.