Aiming at the replication phenomenon of individual components and parts in the selective assembly of complex mechanical products,this paper proposes a new evaluation index of solution set:population evenness, an elite reserved strategy of the offspring based on the combination of population evenness and degree of congestion is proposed，in order to improve the multi-objective optimization NSGA2(Non-dominated Sorting Genetic Algorithm-2) algorithm.With assembly qualified rate and assembly precision as the quality evaluation index, a multi-objective optimization model for selective assembly is established. Use the nearest neighbor search operator to overcome the deficiencies of the local search capability of the NSGA2 algorithm. Taking the assembly of a certain type of aircraft piston engines as an example, the optimization result is represented by the Pareto boundary set. The result shows that the diversity and astringency of the non-dominated solution sets (non-dominated solution sets) are obtained after the algorithm is improved.