Optimal train control via switched system dynamic optimization

This paper considers an optimal train control problem with two challenging, non-standard constraints: a speed constraint that is piecewise-constant with respect to the train's position, and control constraints that are non-smooth functions of the train's speed. We formulate this problem as...

Full description

Bibliographic Details
Main Authors: Zhong, W., Lin, Qun, Loxton, Ryan, Lay Teo, Kok
Format: Journal Article
Language:English
Published: TAYLOR & FRANCIS LTD 2021
Subjects:
Online Access:http://hdl.handle.net/20.500.11937/89486
_version_ 1848765230282178560
author Zhong, W.
Lin, Qun
Loxton, Ryan
Lay Teo, Kok
author_facet Zhong, W.
Lin, Qun
Loxton, Ryan
Lay Teo, Kok
author_sort Zhong, W.
building Curtin Institutional Repository
collection Online Access
description This paper considers an optimal train control problem with two challenging, non-standard constraints: a speed constraint that is piecewise-constant with respect to the train's position, and control constraints that are non-smooth functions of the train's speed. We formulate this problem as an optimal switching control problem in which the mode switching times are decision variables to be optimized, and the track gradient and speed limit in each mode are constant. Then, using control parameterization and time-scaling techniques, we approximate the switching control problem by a finite-dimensional optimization problem, which is still subject to the challenging speed limit constraint (imposed continuously during each mode) and the non-smooth control constraints. We show that the speed constraint can be transformed into a finite number of point constraints. We also show that the non-smooth control constraints can be approximated by a sequence of conventional (smooth) inequality constraints. The resulting approximate problem can be viewed as a nonlinear programming problem and solved using gradient-based optimization algorithms, where the gradients of the cost and constraint functions are computed via the sensitivity method. A case study using data for a real subway line shows that the proposed method yields a realistic optimal control profile without the undesirable control fluctuations that can occur with the pseudospectral method.
first_indexed 2025-11-14T11:31:57Z
format Journal Article
id curtin-20.500.11937-89486
institution Curtin University Malaysia
institution_category Local University
language English
last_indexed 2025-11-14T11:31:57Z
publishDate 2021
publisher TAYLOR & FRANCIS LTD
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-894862023-07-26T05:44:12Z Optimal train control via switched system dynamic optimization Zhong, W. Lin, Qun Loxton, Ryan Lay Teo, Kok Science & Technology Technology Physical Sciences Computer Science, Software Engineering Operations Research & Management Science Mathematics, Applied Computer Science Mathematics Optimal train control switched system control parameterization time-scaling transformation state-dependent control constraint This paper considers an optimal train control problem with two challenging, non-standard constraints: a speed constraint that is piecewise-constant with respect to the train's position, and control constraints that are non-smooth functions of the train's speed. We formulate this problem as an optimal switching control problem in which the mode switching times are decision variables to be optimized, and the track gradient and speed limit in each mode are constant. Then, using control parameterization and time-scaling techniques, we approximate the switching control problem by a finite-dimensional optimization problem, which is still subject to the challenging speed limit constraint (imposed continuously during each mode) and the non-smooth control constraints. We show that the speed constraint can be transformed into a finite number of point constraints. We also show that the non-smooth control constraints can be approximated by a sequence of conventional (smooth) inequality constraints. The resulting approximate problem can be viewed as a nonlinear programming problem and solved using gradient-based optimization algorithms, where the gradients of the cost and constraint functions are computed via the sensitivity method. A case study using data for a real subway line shows that the proposed method yields a realistic optimal control profile without the undesirable control fluctuations that can occur with the pseudospectral method. 2021 Journal Article http://hdl.handle.net/20.500.11937/89486 10.1080/10556788.2019.1604704 English TAYLOR & FRANCIS LTD fulltext
spellingShingle Science & Technology
Technology
Physical Sciences
Computer Science, Software Engineering
Operations Research & Management Science
Mathematics, Applied
Computer Science
Mathematics
Optimal train control
switched system
control parameterization
time-scaling transformation
state-dependent control constraint
Zhong, W.
Lin, Qun
Loxton, Ryan
Lay Teo, Kok
Optimal train control via switched system dynamic optimization
title Optimal train control via switched system dynamic optimization
title_full Optimal train control via switched system dynamic optimization
title_fullStr Optimal train control via switched system dynamic optimization
title_full_unstemmed Optimal train control via switched system dynamic optimization
title_short Optimal train control via switched system dynamic optimization
title_sort optimal train control via switched system dynamic optimization
topic Science & Technology
Technology
Physical Sciences
Computer Science, Software Engineering
Operations Research & Management Science
Mathematics, Applied
Computer Science
Mathematics
Optimal train control
switched system
control parameterization
time-scaling transformation
state-dependent control constraint
url http://hdl.handle.net/20.500.11937/89486