Integrated Machine Learning Framework for Optimizing Residual Material Selection in Environmental Remediation

This project will develop an easy-to-understand and transparent machine learning system that helps identify the best residual materials to clean up environmental contamination, especially in areas affected by mining. Materials such as biochar, zeolite, dolomite, and wood ash are already known to help remove pollutants like heavy metals, phenolic compounds, acid mine drainage, and gases, but current research is scattered and often hard to compare. To address this, the project will bring together information from laboratory tests, pilot experiments, and long-term field studies into one organized dataset. Using this combined data, the intern will build machine learning models that can compare materials and recommend which option works best for a specific contamination scenario. The project will also include simple explanation tools to show why the model makes each recommendation and how confident it is in its predictions. Field data from the Abitibi-Témiscamingue region, where mining and forestry activities produce large amounts of residual materials, will be used to test and refine the system. By the end of the project, the participating institutions will benefit from a practical decision-support tool that improves environmental management, supports sustainable reuse of industrial by-products, and strengthens collaboration between research, industry, and local communities.

Faculty Supervisor:

Flavia Braghiroli

Student:

Partner:

Universidade de São Paulo

Discipline:

Engineering

Sector:

Education

University:

Université du Québec en Abitibi-Témiscamingue

Program: