This work focuses on the design of a PID controller for a propeller pendulum system based on the approximation of a transfer function using a Hammerstein-Wiener neural network (H-W NN). The identification process is performed by obtaining experimental data and the H-W NN model is presented with a reduction approach without loss of accuracy and computational cost under a reduction method. Subsequently, a transfer function based on approximations with the H-W NN is proposed and the Matlab PID tuner is used to obtain the controller values. The experimental data of the PID controller applied to the propeller pendulum and the simulation data are shown.