• Unearthed Sydney 2018, Team Regressive: 1st place and People's Choice Award. From left; Aqeel Akber, Prithvi Reddy, Tom Zheng, Florian Fahrenholz, Mahasen Sooriyabandara.

Throughput optimisation for mines by tuning truck payload and speed

A platform for high-level decision making using deep insights based on real-world data

Working with Evolution’s Cowal Gold Mine we conducted operations research into their management of their fleet of trucks. Given data about their mine and truck operations, we developed a bespoke machine learning model to predict the total time taken for a truck to move a given mass between any two points in the mine. Through careful statistical analysis, we used this model to predict payloads and speed to maximise throughput.

This model is integrated into a user-friendly interface for mining supervisors and dispatchers. We developed a web platform to present the analytics, including a 3D-viewer of the mine pit with route-segments highlighted to show live performance.

This project has fostered an ongoing relationship with Evolution Mining for their data science needs.