Abstract: The influence of encoding mechanism on the convergence of genetic algorithm is analyzed. High carry number encoding has the ability of faster searching to optimization area, compared with low carry number encoding while the population is unstable. Low carry number encoding has the ability of faster searching to optimization value, compared with high carry number encoding while the population is stable. Then, the genetic algorithm based on degressive carry number encoding is proposed. The algorithm can improve the convergence speed of optimization problems, and this advantage is obvious while the parameters are excessive. Simulations validate the conclusion.