Font Size: a A A

Study On Interactive Fuzzy Preference Multi-objective Evolutionary Algorithm

Posted on:2020-09-03Degree:MasterType:Thesis
Country:ChinaCandidate:R N ZhouFull Text:PDF
GTID:2428330602450607Subject:Engineering
Abstract/Summary:PDF Full Text Request
Helping the decision maker(DM)in finding the most preferred solution is the main purpose of the preference-based multi-objective evolutionary algorithm(MOEA).In recent years,interactive methods that can achieve this goal and have been paid more and more attention.In the interactive preference-based MOEAs,on the one hand,the DM can optimize search by incorporating their preferences into the search process.On the other hand,the DM can adjust the preference information dynamically,and find the preferred solution(s)quickly.The current interactive preference-based MOEAs require the DM to give accurate preference information,however,in practice,it may be difficult for the non-professional DM to precisely state his/her preferences.Thus,this paper proposes the methods based on the decomposition of multi-objective evolutionary algorithm(MOEA/D-DE)framework to study the interactive fuzzy preference MOEA,it is mainly divided into the following three parts: 1.An interactive multi-objective evolutionary algorithm based on fuzzy language preference is proposed.In view of the uncertain preference of the DM for objectives,the algorithm designs a utility function preference model based on fuzzy language.Firstly,the preference information is expressed according to the fuzzy language value provided by the DM,and then the utility function is used to model the preference information.It is more convenient and intuitive to express preference information.The simulation conducted on the standard multi-objective optimization problems and extensive experiments proved that the proposed algorithm is convenient,reasonable and effective.2.An interactive multi-objective evolutionary algorithm based on fuzzy preferences in constrained regions is proposed.This algorithm designs a fuzzy utility function preference model based on constrained regions.The preference model contains two types of preference information: one is that the DM provides a reference point to determine a rough preferred region,and the other is that in the determined preferred region,the DM provides the fuzzy language value to express the preference degree for different objectives.The experimental results shown that the feasibility and effectiveness of the proposed algorithm.3.An interactive multi-objective evolutionary algorithm based on fuzzy preference of variable regions is proposed.In order to allow the DM to fully participate in the evolutionary process,this algorithm designs a fuzzy utility function preference model based on variable regions.According to the different fuzzy preference information given by the DMs,different preference models can be established.At each interaction,the algorithm can change the preference information according to the DM's different preference needs until the satisfactory preferred solution(s)is found.Experimental results shown that the proposed algorithm can flexibly change the preferred region.
Keywords/Search Tags:Multi-objective Optimization, Decomposition, Interactive, Fuzzy Preference, Utility Function
PDF Full Text Request
Related items