Recurrent neural network control of functional electrical stimulation systems

In this paper, a recurrent neural network (RNN) controller is proposed for the application of functional electrical stimulation (FES) system, which is a fast developing technique in the area of rehabilitation engineering. With the proposed scheme, the FES system can obtain a better response speed and an improved robustness against disturbance compared to a PID controlled one. Furthermore, L₂-stability of RNN training algorithm is guaranteed via input-output analysis from the nonlinear system theory. Finally based upon a musculoskeletal model, computer simulations are carried out to verify the effectiveness of the theoretical results.