Electro-Mechanical Actuator (EMA), as the main actuating system of advanced aircraft, has the advantages of small size, high efficiency and light weight. It is main development direction of future aviation actuator. Firstly, the working principle of EMA was introduced, the non-linear model of mechanical transmission structure and permanent magnet synchronous motor was established, and a three-loop EMA control system with voltage feed-forward was constructed. Then, considering the influence of mechanical transmission system on EMA position loop, a radial basis function (RBF) neural network PID controller was constructed for position loop by using the identification ability of neural network. The results show that the control accuracy and response rate have been improved obviously.