Knowledge base: Warsaw University of Technology

Settings and your account

Back

Facility location problems with complex preference models

Paweł Olender

Abstract

The dissertation concerns discrete multicriteria location problems and specially preference models that generate optimal solutions in the sense of outcomes distribution. Such an approach allows to obtain impartial solutions, where the distribution of outcomes is more important than their assignment to specific criteria. It is also possible to provide fairness as equitable efficiency. Analyzed approach is derived from the Ordered Weighted Averaging (OWA) operator. Different OWA and WOWA optimization models are presented and proposed including both general MILP and specific LP formulations for suitable monotone weights. New optimization models for the Weighted Ordered Weighted Averaging (WOWA) operator allow to take into account demand requirements, while preserving the impartiality condition. The performance of different formulations is studied and influence of redundant constraints is analyzed. A heuristic method, so-called Variable Neighborhood Search (VNS), is also developed. The proposed modifications of VNS significantly reduce solution time and simultaneously improve quality of solutions.
Record ID
WUT8659d5e1137e4ea8be4bd5b586a25e29
Diploma type
Doctor of Philosophy
Author
Paweł Olender Paweł Olender,, Undefined Affiliation
Title in Polish
Zagadnienia lokalizacyjne ze złożonymi modelami preferencji
Title in English
Facility location problems with complex preference models
Language
(pl) Polish
Certifying Unit
Faculty of Electronics and Information Technology (FEIT)
Discipline
information science / (technology domain) / (technological sciences)
Status
Finished
Defense Date
11-06-2015
Title date
11-06-2015
Supervisor
Internal reviewers
External reviewers
Andrzej Janiszewski Andrzej Janiszewski,, External affiliation of publication: Politechnika Poznańska
Pages
199
Keywords in English
multicriteria optimization, discrete location problems, ordered operators, outcomes distribution, equitable efficiency, OWA operator, WOWA operator, VNS method
Abstract in English
The dissertation concerns discrete multicriteria location problems and specially preference models that generate optimal solutions in the sense of outcomes distribution. Such an approach allows to obtain impartial solutions, where the distribution of outcomes is more important than their assignment to specific criteria. It is also possible to provide fairness as equitable efficiency. Analyzed approach is derived from the Ordered Weighted Averaging (OWA) operator. Different OWA and WOWA optimization models are presented and proposed including both general MILP and specific LP formulations for suitable monotone weights. New optimization models for the Weighted Ordered Weighted Averaging (WOWA) operator allow to take into account demand requirements, while preserving the impartiality condition. The performance of different formulations is studied and influence of redundant constraints is analyzed. A heuristic method, so-called Variable Neighborhood Search (VNS), is also developed. The proposed modifications of VNS significantly reduce solution time and simultaneously improve quality of solutions.
Thesis file

Uniform Resource Identifier
https://repo.pw.edu.pl/info/phd/WUT8659d5e1137e4ea8be4bd5b586a25e29/
URN
urn:pw-repo:WUT8659d5e1137e4ea8be4bd5b586a25e29

Confirmation
Are you sure?
Report incorrect data on this page