Predicting energy consumption SAG mills through Bayesian generalized linear model and random forest.

Area of Science:

• Mining Engineering
• Energy Management
• Data Science in Mining

Background:

• The global mining industry is a significant energy consumer, with milling processes being the most intensive.
• Energy consumption in mining is projected to rise, necessitating efficient prediction and management strategies.
• Several factors, including rock properties and equipment specifics, influence milling energy usage.

Purpose of the Study:

• To develop and evaluate predictive models for SAG (Semi-Autogenous Grinding) mill energy consumption.
• To compare the performance and insights offered by Random Forests and Generalized Linear Models (GLM) for energy prediction.
• To identify key variables influencing SAG mill energy consumption in a copper mining operation.

Main Methods:

• Application of Random Forests and Generalized Linear Models (GLM) for energy consumption forecasting.
• Case study utilizing a South American copper mine dataset.
• Analysis of model accuracy, feature importance, and uncertainty quantification.

Main Results:

• Both Random Forests and GLM successfully predicted SAG mill energy consumption.
• Random Forest achieved high prediction accuracy (95%) but limited explanatory power (R2 at 50%).
• GLM provided feature importances and probability distributions, offering a comprehensive view of energy consumption scenarios.

Conclusions:

• Random Forests and GLM are effective tools for predicting mining energy consumption.
• GLM offers superior interpretability, detailing variable impacts and potential energy consumption ranges.
• Accurate energy consumption prediction is crucial for optimizing mining operations and managing energy resources effectively.