Abstract: The dependence among decision variables in complex optimization problems leads to the appearance of a large number of local optimal solutions in the fitness landscape of functions, which are difficult to be solved by classical evolutionary algorithms. In this paper, a constructive learning success-history based adaptive differential evolution (CLSHADE) algorithm is proposed to solve partially separable function optimization problems. Firstly, CLSHADE uses the differential grouping technology (DG) to reduce the complexity of a complex problem by dividing it into multiple sub-problems. Then, a constructive learning strategy, based on the grouping structure, is designed. It learns from the constructed optimal solution in a certain probability to guide the search direction and improve the search performance of the CLSHADE. The experimental results on the partially separable function of CEC 2017 demonstrate the effectiveness of the CLSHADE.