Machine Learning and Soft Sensing for Earth-Derived Material Processes in Industry

The field of industrial production is increasingly dependent on materials whose origins and behaviors are intricately tied to geological processes and mineralogical characteristics. As the need for sustainable and efficient manufacturing intensifies, data-driven tools such as machine learning (ML) and soft sensors are emerging as transformative technologies, offering granular, real-time perspectives on the complex thermal, chemical, and mechanical transformations occurring during the extraction and processing of Earth-derived materials. Despite advancements, the challenge remains to reliably interpret rapid and nonlinear material behaviors that conventional monitoring approaches cannot adequately capture. Recent studies demonstrate promising outcomes: ML models now reveal latent relationships between material composition, structure, and industrial performance, while soft sensors infer critical process variables from indirect or limited measurements, and hybrid techniques combine domain knowledge with predictive power. These innovations can enhance forecasting accuracy, optimize material usage, and minimize environmental harm at every stage of the material lifecycle. Nevertheless, their integration in the context of Earth materials and geoscience-driven industries is still nascent, with interdisciplinary communication and dedicated applications underdeveloped in mainstream Earth science literature.This Research Topic aims to foreground data-driven strategies—particularly ML and soft sensing—that improve our ability to monitor, understand, and optimize the behavior of geological and mineral-based materials in real-time industrial contexts. It seeks to connect the expertise of geoscientists, material scientists, process engineers, and data scientists to crack open fundamental and applied questions: How do digital tools deepen our understanding of material evolution during industrial processing? In what ways can these approaches promote resource efficiency, sustainability, and superior product performance? By encouraging submissions that bridge knowledge of natural mineral properties with digital process control, the topic aspires to expand the frontiers of sustainable innovation.Within the boundaries of integrating data-driven methods with industrial use of geological materials, this research topic focuses on developments that advance real-time monitoring and optimization across extraction, preparation, and transformation stages. At the same time, it excludes purely theoretical machine learning studies with no link to Earth-derived raw materials or industrial application. To guide submissions, we welcome articles addressing the following themes:o Development and validation of soft sensors for real-time tracking of mineralogical, chemical, or microstructural transformationso ML-based prediction of properties, durability, and performance of Earth-derived or mineral-based systemso Data-driven optimization of thermal, chemical, and mechanical processing routes for geological raw materials and industrial by-productso Integration of ML with environmental indicators, emissions monitoring, and sustainability assessments in resource extraction and processingo Case studies demonstrating improved efficiency, emissions reduction, or product quality via digital toolso Hybrid modeling approaches that merge geoscientific understanding with ML-based predictiono Interpretability and explainability of ML applications in Earth materials science and industrial processesAppendix: We welcome original research articles, reviews, systematic case studies, and methodological papers. Topic editors: Milica Vasić, Paul Awoyera, Ivanka Netinger Grubeša Source: https://www.frontiersin.org/research-topics/80471/machine-learning-and-soft-sensing-for-earth-derived-material-processes-in-industry