To improve the performance of NSGA-II-DE solving the optimization problem with the complex Pareto solution set, the defection of the crowding distance formula and crowding distance mechanism in the NSGA-II is analyzed. Taking NSGA-II-DE as the basic frame, the crowding-distance sorting method is changed into the two-dimensional information ordering strategy of including angle and pseudo radius. The intervention of orders of magnitude threshold is joined in the crowding distance mechanism, and the improved algorithm(2D-Thr) is proposed, which is compared with NSGA-II-DE, NSGA-II and MACPSO by using the quantitative calculation of three evaluation indexes: the degree of diversity, convergence and spacing. The simulation results show that the improved algorithm not only inherits the excellent convergence of the original algorithm, but also improves the distribution of the Pareto front.