Grinding circuits account for nearly 50% of a mine’s total energy consumption. Yet most control systems operate under the implicit assumption that the plant dynamics remain constant. In reality, they do not.
As mill liners progressively wear, the breakage environment changes. Power draw, throughput, and efficiency shift over time. Operators compensate manually, but traditional control architectures rarely embed liner wear explicitly into their optimisation logic.
Samantha Krause, a Bachelor/Master of Engineering student undertaking her industry placement at Sustainable Minerals Institute's Julius Kruttschnitt Mineral Research Centre (JKMRC)under the supervision of Dr Christian Zuluaga Bedoya and Dr Peter Bohm, is addressing this gap.
Industry Placement Project
Her project integrates empirical liner wear prediction models with Deep Reinforcement Learning (DRL) in SAG mill, Ball mill, and Crusher circuits, commonly known as the SABC circuit in the industry. By interfacing an existing dynamic circuit model (MATLAB/Simulink) with a custom Python-based DRL environment, she has developed a supervisory optimisation layer capable of adapting set-points as liner conditions evolve.
The DRL agent operates on continuous-state inputs, including power draw, throughput, and liner age, and optimises continuous control actions. Reward structures are designed to balance power efficiency, operational stability, and liner life.
Advanced Machine Learning for existing control architectures
The result is not just better control tuning. It is a shift toward wear-aware decision systems. where digital twins, data-driven models, and reinforcement learning work together to adapt plant operation over the full liner lifecycle.
This work demonstrates how advanced machine learning can be embedded into existing control architectures to improve throughput, reduce energy intensity, extend equipment life, and lower carbon footprint.
At JKMRC, industry placement students are not observers. They contribute to the next generation of adaptive, decision-centric mineral processing systems.
Find out more about student research opportunities here: https://employability.uq.edu.au/summer-winter-research