Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.

  

Learning Control for Functional Electrical Stimulation Excited Quadriceps Knee Angle Control

Author(s):

I. Abou-Zeid
Conference/Journal:

Master Thesis, FS13 (10247)
Abstract:

Restoration of motor function with people that suffer from Spinal Cord Injury (SCI) is most commonly accomplished by Functional Electrical Stimulation (FES). Despite the high number of existing solutions only few are available on the market for their impracticability or unsatisfactory tracking precision and disturbance rejection by reasons of the time-varying nature of muscle excitement. This thesis aimed at exploring the potential of Iterative Learning Control (ILC) to control FES incited systems by means of a simulation study on the example of quadriceps incited knee angle control. A simple anticipatory ILC was shown to not guarantee robust convergence for the used Single-Input Single- Output (SISO) model, which may change when using a more complex model including co-contraction of muscles. A Norm-Optimal ILC consisting of a 2-step approach was implemented and showed low computational effort and fast convergence with low tracking error sensitivity on noise, repetitive disturbance and initial state error. Parameter sensitivity was critical with respect to the input gain, which in practice occurs due to muscle fatigue and a change in the implementation by means of changes in the model correction is suggested to overcome this critical drawback and further changes are suggested to better overall performance. A Direct FedbacK (DFK) scheme which generates controller directly from measurement data was implemented using different measurement sets. Measurements for only the relevant part of the state space lead to better performing controllers in ideal conditions, yet they are more prone to model mismatch. Random measurements deliver less precise yet acceptable tracking performance under ideal conditions but are less prone to parameter mismatch, especially with respect to input gain. Since full state information is needed the analysis should be extended to observers. A combination between Norm-Optimal ILC and DFK is suggested as an extension.

Supervisors: Robert Nguyen, Marko Tanaskovic, Manfred Morari

Year:

2015
Type of Publication:

(12)Diploma/Master Thesis
Supervisor:

M. Morari

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2015:IFA_5388
}
Permanent link