Decision Support in Algorithm Selection for Generic Optimisation

Research output: Chapter in Book/Report/Conference proceedingConference contribution

130 Downloads (Pure)

Abstract

This paper presents the development of an algorithm-selection framework supported by a new intuitive user interface for the generic optimisation tool, GenOpt. The framework consists of an algorithm-selection flowchart to help identify relevant algorithms depending on the nature of the problem, followed by an algorithm-selection matrix which evaluates the algorithms’ suitability based on the user requirements. The algorithm selection framework acts as a decision support system to allow the user to select the most appropriate and effective optimisation algorithm for a given problem. Such a procedure improves decision-making, limits the algorithm selection errors and helps the user to achieve solutions closer to the Pareto optimum. The selection framework is supported by a user interface, developed in C++ and compatible with GenOpt, that allows users who do not have prior coding knowledge to use GenOpt successfully. The developed interface presents the user with the most relevant optimisation algorithms from those available in the programme. It allows the user to easily modify algorithmic variables in a user-friendly environment. The novelty of the approach is reflected in the built-in knowledge and intelligence in the pre-selection of optimisation algorithms, which are tailored to specific user-defined problems. This, consequently, improves the overall optimisation results by allowing the user to better understand the optimisation algorithm and its variables.
Original languageEnglish
Title of host publicationProceedings of the 5th IBPSA-England Conference on Building Simulation and Optimization (Virtual)
EditorsBianca Howard, Argyris Oraiopoulos, Eleonora Brembilla
Place of PublicationLoughborough, UK
PublisherIBPSA England
Number of pages8
Publication statusPublished - 21 Sep 2020
EventBuilding Simulation and Optimization 2020: IBPSA England's First Virtual Conference - Loughborough University, Loughborough, United Kingdom
Duration: 21 Sep 202022 Sep 2020
https://www.bso2020.org/

Conference

ConferenceBuilding Simulation and Optimization 2020
Abbreviated titleIBPSA
CountryUnited Kingdom
CityLoughborough
Period21/09/2022/09/20
Internet address

Fingerprint Dive into the research topics of 'Decision Support in Algorithm Selection for Generic Optimisation'. Together they form a unique fingerprint.

Cite this