To a given double pendulum system whose parameters are unknown and whose output is limited, this paper relates two methods driving the second arm of the system from its natural pendant position to the right-up position. In this system, one-swing movement refers to when the second arm starts and ends with speed 0. The swing-up movement can be considered to be the combination of the one-swing movement. First, giving a required output over a finite time interval for one-swing movement, we can obtain a required input by using the learning control method proposed in this paper. Repeating this operation, the required inout to swing the arm to the inverted position may be constructed. However, the following control method may be effective if the required input can be obtained by system simulation even if the actual system parameters are different from those of simulation. In any case, the effectiveness of both is assured by experiment results.