Abstract: It is known that the convergence speed of Genetic Algorithm (GA) is slow. The paper proposes the methods of improving the speed of GA, which are methods remaining the current best solution during mutation process, searching different regions, changing the cluster length that represents the different variables (near the position of the best solution, searching step length.is got small). First, Computer simulations are performed by the different methods. Second, another computer simulation is performed by combining above three methods. The latter simulation shows that this method has greatly improved the speed of GA and can be applied in some real control processes.