In this paper, based on correlativity theory, a kind of multi-objective evolutionary algorithm is put forward. First, the best solution of every objective among the mufti-objectives is obtained and they are regarded on as the referenced vector. Second, the correlativity index between every individual and the referenced vector is solved and the correlativity index is acted as fitness of the individual. Moreover, the pareto optimal sets are solved by means of adaptive genetic algorithm. The variety of population is kept by means of adaptive probability of crossover and mutation. At last, the algorithm is used to optimize the design parameters of cylinder helical compression spring. Simulation examples show the effectiveness of the approach proposed.