The algorithms, developed by DINGO and the Queensland University of Technology were nominated for the Innovation category of the AMM Awards in association with Komatsu.
They have been applied to DINGO’s Trakka predictive maintenance software to deliver better intelligence and results to customers.
The algorithms’ development has been driven by the exponential growth in condition monitoring data, largely driven by sensor data.
That has created the need for technologies that can augment human capability and existing analytical platforms.
The optimal solutions is artificial intelligence, however, given DINGO has little experience in that area it collaborated with data scientists and statistical mathematics experts from QUT through the Advanced Queensland Knowledge Program.
That program has received a grant from the Queensland government that funded the employment of a QUT PhD graduate in mathematical statistics.
That graduate was embedded in the DINGO business.
That person received weekly mentoring from QUT data science experts who helped identify and apply theories to machine learning problems.
DINGO maintenance experts worked with the QUT graduate to identify variables from existing data sets and test and modify open source machine learning algorithms.
An iterative research and application process produced a series of machine learning models that enabled sophisticated predictive analytics.
This intelligence is helping miners identify potential issues earlier and get improved machine availability and performance.
Those machine learning models have helped DINGO progress through an engagement with a leading miner that is exploring solutions in the predictive maintenance space.
Noel Dyson │ Australia’s Mining Monthly