Piecewise-Linear Switching Surfaces for Minimum Time Control.
Procedures for designing piecewise-linear (PWL) switching surfaces for minimum time control are investigated. The approach is to generate the switching function using the instantaneous values of the states as inputs to easily realized PWL function generators. The design problem is then one of fittin...
Main Author: | |
---|---|
Other Authors: | |
Language: | en_US |
Published: |
Monterey, California. U.S. Naval Postgraduate School
2013
|
Online Access: | http://hdl.handle.net/10945/28450 |
id |
ndltd-nps.edu-oai-calhoun.nps.edu-10945-28450 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-nps.edu-oai-calhoun.nps.edu-10945-284502014-11-27T16:17:26Z Piecewise-Linear Switching Surfaces for Minimum Time Control. Redderson, Roy Henry. NA NA Procedures for designing piecewise-linear (PWL) switching surfaces for minimum time control are investigated. The approach is to generate the switching function using the instantaneous values of the states as inputs to easily realized PWL function generators. The design problem is then one of fitting the state space PWL surface to points defining the optimal surface. Several performance indices related to minimum-time are defined and the resulting parameter surfaces for different initial condition distributions are studied. However, because of local minima, a successful search procedure could not be found. Therefore, a heuristic method of least squares fitting is used. A combination gradient and generalized Newton-Raphson search method is employed to obtain values for the PWL parameters. Several least squares fit methods are applied to a second order problem and the results compared using response time performance to a uniform grid of initial conditions. Very close to optimal performance is achieved using a delay time design procedure. 2013-02-15T23:33:18Z 2013-02-15T23:33:18Z 1966 Thesis http://hdl.handle.net/10945/28450 en_US Monterey, California. U.S. Naval Postgraduate School |
collection |
NDLTD |
language |
en_US |
sources |
NDLTD |
description |
Procedures for designing piecewise-linear (PWL) switching surfaces for minimum time control are investigated. The approach is to generate the switching function using the instantaneous values of the states as inputs to easily realized PWL function generators. The design problem is then one of fitting the state space PWL surface to points defining the optimal surface. Several performance indices related to minimum-time are defined and the resulting parameter surfaces for different initial condition distributions are studied. However, because of local minima, a successful search procedure could not be found. Therefore, a heuristic method of least squares fitting is used. A combination gradient and generalized Newton-Raphson search method is employed to obtain values for the PWL parameters. Several least squares fit methods are applied to a second order problem and the results compared using response time performance to a uniform grid of initial conditions. Very close to optimal performance is achieved using a delay time design procedure. |
author2 |
NA |
author_facet |
NA Redderson, Roy Henry. |
author |
Redderson, Roy Henry. |
spellingShingle |
Redderson, Roy Henry. Piecewise-Linear Switching Surfaces for Minimum Time Control. |
author_sort |
Redderson, Roy Henry. |
title |
Piecewise-Linear Switching Surfaces for Minimum Time Control. |
title_short |
Piecewise-Linear Switching Surfaces for Minimum Time Control. |
title_full |
Piecewise-Linear Switching Surfaces for Minimum Time Control. |
title_fullStr |
Piecewise-Linear Switching Surfaces for Minimum Time Control. |
title_full_unstemmed |
Piecewise-Linear Switching Surfaces for Minimum Time Control. |
title_sort |
piecewise-linear switching surfaces for minimum time control. |
publisher |
Monterey, California. U.S. Naval Postgraduate School |
publishDate |
2013 |
url |
http://hdl.handle.net/10945/28450 |
work_keys_str_mv |
AT reddersonroyhenry piecewiselinearswitchingsurfacesforminimumtimecontrol |
_version_ |
1716725021911547904 |