Calibration of Genetic Algorithm Parameters for Mining-Related Optimization Problems

Genetic algorithms (GA) are widely used to solve engineering optimization problems. The quality and performance of the solution generated strongly depend on the selection of the GA parameter values (crossover and mutation rates and population size). We propose an approach based on full factorial and...

Full description

Saved in:
Bibliographic Details
Published in:Natural resources research (New York, N.Y.) Vol. 28; no. 2; pp. 443 - 456
Main Authors: Villalba Matamoros, Martha E., Kumral, Mustafa
Format: Journal Article
Language:English
Published: New York Springer US 01.04.2019
Springer Nature B.V
Subjects:
Algorithms
Calibration
Chemistry and Earth Sciences
Computer Science
Computing time
Crossovers
Earth and Environmental Science
Earth Sciences
Fossil Fuels (incl. Carbon Capture)
Genetic algorithms
Geography
Heuristic methods
Mathematical Modeling and Industrial Mathematics
Mineral Resources
Mutation
Mutation rates
Objective function
Optimization
Original Paper
Parameters
Physics
Population number
Response surface methodology
Statistics for Engineering
Sustainable Development
Underground mines
Underground mine planning
Genetic algorithms (GA)
Stope layout optimization
Response surface methodology
GA parameters
ISSN:1520-7439, 1573-8981
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Genetic algorithms (GA) are widely used to solve engineering optimization problems. The quality and performance of the solution generated strongly depend on the selection of the GA parameter values (crossover and mutation rates and population size). We propose an approach based on full factorial and response surface methodology experimental designs to calibrate GA parameters such that the objective function is maximized/minimized and the relative importance of the parameters is quantified. The approach was tested by applying it to stope optimization of underground mines, where profit can vary ± 7% based solely on GA parameters. Results showed that: (1) a larger population size did not always increase solution time; (2) solution time was positively related to crossover and mutation rates; and (3) simultaneous analysis of solution time and profit illustrated the trade-off between acceptable computing time and profit desirability through GA parameter selection. This approach can be used to calibrate parameters of other metaheuristics.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1520-7439
1573-8981
DOI:10.1007/s11053-018-9395-2