Item – Thèses Canada

​
Numéro d'OCLC
1117446810
Lien(s) vers le texte intégral
Exemplaire de BAC
Auteur
Vargas Calixto, Carlos,
Titre
Temporal expansion and nonlinear parameter varying methods for the identification of nonlinear time-varying ankle dynamic stiffness
Diplôme
M. Eng. -- McGill University, 2019
Éditeur
[Montreal] : McGill University Libraries, [2019]
Description
1 online resource
Notes
Thesis supervisor: Robert E Kearney (Internal/Supervisor).
Includes bibliographical references.
Résumé
"The identification of a comprehensive model of dynamic joint stiffness must deal with its time-varying and nonlinear behavior. Our laboratory has developed two new methods designed to deal with these problems. The temporal expansion (TE) method assumes that the system parameters vary explicitly with time and identifies these variations as functions of time. The nonlinear parameter varying (NPV) method assumes that the parameters vary with some known scheduling variable (SV), which may vary with time, and identifies how these parameters vary with the SV. This thesis compares the performance of the two methods when applied to data generated by simulating a realistic model of ankle dynamic joint stiffness. The results showed that both methods could successfully track the changes in stiffness dynamics. For both methods, the average %VAF increased when the signal to noise ratio or experiment length increased. However, the %VAF for the NPV method dropped at positions that were underrepresented in the SV distribution. For the TE method, the %VAF was constant throughout the cycle, except at the beginning and end of the record where it dropped due to transients. In terms of the estimated model, the NPV results were more accurate for intrinsic and reflex dynamics, whereas TE estimated the reflex static nonlinearity better. Also, PRALDS inputs generated the best results, whereas PRBS inputs caused convergence problems. Finally, this thesis demonstrates that TE results can be used to choose what SV to use with NPV, by showing how the system parameters covary with other variables."--
Autre lien(s)
digitool.Library.McGill.CA
escholarship.mcgill.ca
escholarship.mcgill.ca
Sujet
Political Science