Abstract:
Real-world functional electrical stimulation (FES) encounters nonlinear effects of fatigue and
time delay that cause the FES controller to under-perform or sometimes fail. Nonlinearities
cause the system to change, but it is not possible to re-tune the controller once its
parameters are set. System representation using an observer can allow the patient’s knee
extension to be represented in a numerical computation algorithm and can exist, run, or be
executed in an embedded system. This enables the closed-loop controller to be tuned to the
system being imitated. The second-order black box model can be matched to the Veltink
model to represent the knee extension in which the model is transformed into a linear
differential equation, and then into a physical-based model. The numerical computation
using Taylor series is then used to convert the physics-based model using a computational
algorithm that represents the knee extension system in a discrete-time linear system. Prior
to the conversion of the numerical model, certain parameters need to be extracted from the
actual system response of the patient knee angle trajectory upon receiving an open-loop
signal from FES. This paper focuses on the control algorithm technique for extracting the
second-order black box model parameters matched the actual knee extension system
response. MATLAB/Simulink is used to test the parameter extraction algorithms. The results
indicate that the extraction algorithms succeeded in extracting the actual system
parameters that are similar to the ones obtained by extracting the parameters representing
the knee extension model. © 2022, The Author(s), under exclusive license to Springer
Nature Switzerland AG.
