Algorithms for design optimization of chemistry of hard magnetic alloys using experimental data

Abstract

A multi-dimensional random number generation algorithm was used to distribute chemical concentrations of each of the alloying elements in the candidate alloys as uniformly as possible while maintaining the prescribed bounds on the minimum and maximum allowable values for the concentration of each of the alloying elements. The generated candidate alloy compositions were then examined for phase equilibria and associated magnetic properties using a thermodynamic database in the desired temperature range. These initial candidate alloys were manufactured, synthesized and tested for desired properties. Then, the experimentally obtained values of the properties were fitted with a multi-dimensional response surface. The desired properties were treated as objectives and were extremized simultaneously by utilizing a multi-objective optimization algorithm that optimized the concentrations of each of the alloying elements. This task was also performed by another conceptually different response surface and optimization algorithm for double-checking the results. A few of the best predicted Pareto optimal alloy compositions were then manufactured, synthesized and tested to evaluate their macroscopic properties. Several of these Pareto optimized alloys outperformed most of the candidate alloys on most of the objectives. This proves the efficacy of the combined meta-modeling and experimental approach in design optimization of the alloys. A sensitivity analysis of each of the alloying elements was also performed to determine which of the alloying elements contributes the least to the desired macroscopic properties of the alloy. These elements can then be replaced with other candidate alloying elements such as not-so-rare earth elements

Design of High Temperature Ti–Al–Cr–V Alloys for Maximum Thermodynamic Stability Using Self-Organizing Maps

“…SOMs can be considered as a nonlinear generalization of principal component analysis (PCA), an unsupervised machine learning method [20,22]. Recent studies demonstrated the advantage of using SOM over PCA [10,14]. Most importantly, SOMs have been successfully used for feature extraction of scarce datasets (sample size of about 40), whereas conventional neural networks require very large training datasets.…”

“…SOMs can be considered as a nonlinear generalization of principal component analysis (PCA), an unsupervised machine learning method [20,22]. Recent studies demonstrated the advantage of using SOM over PCA [10,14].…”

“…In order to significantly reduce the need for experimentation, reduce the total amount of time spent designing new alloys, and eliminate the subjectivity factor of the designer, we have developed a computational design methodology using multiobjective evolutionary optimization algorithms and response surfaces with periodic experimental verifications [6][7][8][9][10][11] applicable to arbitrary alloys.…”

Advanced Heterogeneous Fenton Treatment of Coalbed Methane-Produced Water Containing Fracturing Fluid

“…Coalbed methane is an important source of natural gas that is produced during coal formation and found in coal seams in an adsorbed state [1,2]. Coalbed methane is generally produced through drainage and pressure lowering [3]. In this process, coalbed methane-produced water is also produced.…”

On the evolution of Cu-Ni-rich bridges of Alnico alloys with tempering

“…Spinodal decomposition occurs during thermomagnetic treatment, and the temperature and time of this step have been studied to optimize the size and shape of the α1 phase formed [1,[3][4][5][6][7][8][9][10][11][12][13][14][15].…”