Dolomite-Derived Grain-Oriented Silicon Steel-Grade Magnesium Oxide: Process Optimization and Machine-Learning-Assisted Purity Prediction

Description

This dataset supports the manuscript entitled “Dolomite-Derived Grain-Oriented Silicon Steel-Grade Magnesium Oxide: Process Optimization and Machine-Learning-Assisted Purity Prediction”. The dataset includes experimental process data, characterization data, machine-learning modeling files, and validation data related to the preparation and purity prediction of dolomite-derived grain-oriented silicon steel-grade magnesium oxide. The main experimental dataset contains 305 samples and 28 process parameters, including calcination, hydration, carbonation, complexation-assisted purification, pyrolysis, and precursor calcination conditions. MgO purity was used as the target variable for machine-learning modeling. The deposited files also include variable definitions, source data for the figures, new-process validation data, Python scripts for feature engineering and model training, SHAP and permutation-importance analysis, error analysis, and GUI-based MgO purity prediction. This file provides the variable dictionary for the experimental and machine-learning datasets used in this study. It includes definitions, units, process stages, data types, model roles, and descriptions of the raw process variables, engineered features, target variable, and prediction-output variables related to MgO purity prediction. These data and code can be used to reproduce the main machine-learning workflow, evaluate the final ExtraTrees–LightGBM–KNN non-negative weighted ensemble model, and support model-guided screening of new MgO preparation conditions.

Institutions

• Taiyuan University of Technology

  Shanxi, Taiyuan

Categories

Licence