Comparative Study of Multi/Many-Objective Evolutionary Algorithms on Hot Rolling Application
Lists
Only you see this
Abstract
Handling multiple number of objectives in industrial optimizationOptimization problems is a regular affair. The journey of development of evolutionary algorithmsEvolutionary algorithm for handling such problems occurred in two phases. In the first phase, multi-objective optimizationMulti-objective optimization algorithms are developed that worked quite satisfactorily while finding ParetoPareto set of solutions for two to three objectives. However, their success rates for finding the ParetoPareto optimal solutions for higher number of objectives were limited which triggered the development of different sets of evolutionary algorithmsEvolutionary algorithm under the name of many-objective optimizationOptimization algorithms. In this work, we intend to compare the performance of these two classes of algorithms for an industrial hot rollingHot rolling operation from a real-life steel plant. Several process, chemistryChemistry parameter and geometry related parametersGeometry parameter are modelled to yield different mechanical properties such as % elongationElongation , ultimate tensile strengthUltimate tensile Strength and yield strengthYield strength of final hot rolled steel product through data-based techniques such as artificial neural networks (ANN) Artificial Neural Network (ANN) . Using this ANN model, the mechanical properties are maximized to obtain the ParetoPareto trade-off solutions using both non-dominated sorting genetic algorithms II (NSGA-II) Non-Dominated Sorting Genetic Algorithm (NSGA-II) and many-objective evolutionary algorithm decomposition and dominance (MOEA/DD)Many-Objective Evolutionary Algorithm Decomposition and Dominance (MOEA/DD) and their solutions are compared using a suitable metric for identifying the extent of convergence and diversity. This kind of ParetoPareto set provides a designer with ample of alternatives before choosing a solution for final implementation.
BibTeX
@inproceedings{mittal2019comparative,
author = {Prateek Mittal and Affan Malik and Itishree Mohanty and Kishalay Mitra},
title = {{Comparative Study of Multi/Many-Objective Evolutionary Algorithms on Hot Rolling Application}},
booktitle = {Optimization in Industry},
year = {2019},
url = {https://openreview.net/forum?id=VbA0FCcuZv},
}Loading related content...