By introducing Lamarckian evolutionism, an improved differential evolution algorithm based on the gridded Lamarckian learning mechanism(DE-GLam) is proposed. Under a distributed search framework set by mesh generation mechanism, this algorithm integrates the cell optimum protection mechanism, learning step mechanism, solution space mechanism and directive variation mechanism to form the Lamarck learning mode. The simulation results show that the DE-GLam algorithm not only fully exerts the local search ability of Lamarckian learning mechanism, but also effectively avoids premature convergence, and the solving precision is superior to other comparison algorithms. The validity of the proposed method is illustrated by the optimal power flow calculation.