Modern data analytics approach to predict creep of high-temperature alloys

Abstract

A breakthrough in alloy design often requires comprehensive understanding in complex multicomponent/multi-phase systems to generate novel material hypotheses. We introduce a modern data analytics workflow that leverages high-quality experimental data augmented with advanced features obtained from high-fidelity models. Herein, we use an example of a consistently-measured creep dataset of developmental high-temperature alloy combined with scientific alloy features populated from a high-throughput computational thermodynamic approach. Extensive correlation analyses provide ranking insights for most impactful alloy features for creep resistance, evaluated from a large set of candidate features suggested by domain experts. We also show that we can accurately train machine learning models by integrating high-ranking features obtained from correlation analyses. The demonstrated approach can be extended beyond incorporating thermodynamic features, with input from domain experts used to compile lists of features from other alloy physics, such as diffusion kinetics and microstructure evolution.

Visual Analysis of Odor Interaction Based on Support Vector Regression Method

“…The chemical concentration value of each substance was used as the input variables of the SVR model, and their mixture's olfactory measured odor intensity value was the model's target output variable. The grid search scheme was used to search the model's hyperparameters (the RBF kernel was used, hyperparameters C and γ were optimized), and the 10-fold cross-validation was used to evaluate the prediction ability of a model with certain hyperparameters [32,33]. Based on an optimized SVR model, the odor intensity value of a mixture can be directly predicted after inputting its corresponding chemical composition to the model.…”

Machine learning for material characterization with an application for predicting mechanical properties

“…In the study of Shin et al [139], the five different ML models random forest (RF), linear regression, k ‐nearest neighbor, kernel ridge, Bayesian ridge are applied for the prediction of Larson‐Miller parameters which represent the creep behavior. From a wide range of available features (466), relevant ones are selected with optimization approaches and different set‐ups of features and models evaluated.…”

Recent progress on discovery and properties prediction of energy materials: Simple machine learning meets complex quantum chemistry

“…At present, ML has been applied to the design and optimization of energy materials, including high-temperature alloys [88], perovskites for solar cells [89], lithium-ion battery materials [90,91], catalytic materials [92], phosphors [93], carbon-based supercapacitors [94], and van der Waals (vdW) heterostructures [95] etc. In addition to accelerating the development of energy materials, ML can also optimize the QC calculation methods.…”