![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/73476FEE-4809-45F6-B010-F4BEF8EC6554 | ||
| Year | 2022 | ||
| Type of Item | Doctoral Dissertation | ||
| License |
|
||
| Bibliographic Citation | Konstantinos Zervoudakis, "Multi-objective algorithms for optimal product line design", Doctoral Dissertation, School of Production Engineering and Management, Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.91332 | ||
| Appears in Collections |
Introducing new products has an important role in sustainability and profitability of a firm. Product Line Design (PLD) is a key decision area that product managers have to deal with in the early stages of product development, to estimate the potential success of a product. Even though several objectives may be simultaneously pursued during the product configuration process, most reported studies have focused on single-objective optimization. In this research, the multi-objective PLD (MOPLD) problem is addressed, by taking into account more than one objectives, to provide product managers with a better tradeoff among them, using 23 variants of seven state-of-the-art metaheuristics. The seven main multi-objective metaheuristics used in this research are Genetic Algorithms (GAs), Particle Swarm Optimization (PSO), Firefly Algorithm (FA), Differential Evolution (DE), Grey Wolf Optimizer (GWO), Teaching-Learning Based Optimization (TLBO) and Mayfly Algorithm (MA). Those seven multi-objective algorithms are fully adapted to the MOPLD problem, using popular diversity controlling operators, as well as using an extended to multi-objective optimization Fuzzy-Self-Tuning (FST) process. The purpose of the FST process, is to help the algorithms overcome specific difficulties when performing on different datasets, using the same parameter settings, by calculating the settings of parameters independently for each individual during the optimization process. The 23 variants are compared with each other, through popular performance metrics when it comes to multi-objective optimization, using two types of data sets, under five different MOPLD scenarios, without knowing the specific number of products. Moreover, factors affecting the performance of optimizers are investigated using statistical analysis. Finally, a multi-criteria decision analysis method is used to rank solutions according to the needs of product managers, and an attempt to estimate the possible moves of the competitors, is made.
Copyright © DIAS 2013
