Abstract: With the interpolation principle of high precision multi-layered neural network, a closed-loop control method for swing-up process of circular rail inversed pendulum is investigated based on the optimized open-loop control law. Using the calculating results of state and control variables in open-loop process as standard samples of input and output, a feed-forward neural network is established and a feedback control system for swing-up is closed by the network. Through simulation and practical control, it is obvious that the control method is feasible and efficient. It also provides a new hint for a nonlinear closed control system.