There exists a class of nonlinear periodic disturbance in motor system. This nonlinear disturbance relates to some positions of system, and changes periodic by position. The research presents the system model with position dependent periodic disturbance, supposes to satisfy continuous Lipschitz condition, and derives the condition of position dependent period disturbance translating into time dependent periodic error. A novel iterative learning control method is proposed for attenuation multi-period error of system, which constructs learning operator by error convergence analytic, proposes the condition of compensate multi-period steady-state error, proposes and proves that it accelerates the condition of converging learning operator. The research uses simulation by three-axis test turntable. The simulation results verify that the method can compensate successfully period steady-stat error in motor system, and also proves this method can improve the control precision.