Braden Thorne University of Western Australia

Braden Thorne

PhD Student

Theme 2


Virtual - Researchers Catch-up

Predictive maintenance of any asset hinges on our ability to accurately and frequently measure how said asset is changing throughout operation. This can quickly become a problem if the asset is cumbersome, not readily accessible or vital to operation. The sizer barrels inside of ore crushers manage to tick all of these boxes, so alternative ways for measuring the remaining useful life of these assets need to be considered. Currently, in our context, this is handled with 3D scanning technology, however the time and effort required to get accurate readings means that measurements are often infrequent. 

In this talk Braden discussed computer vision techniques for analysing sizer barrels from CCTV video data, an alternative approach to 3D scans. He outlined the process involved, the techniques used and their requirements, as well as the advantages and disadvantages in comparison to the current method used.  Braden also presented an example output from the computer vision process discussing  how this may be used for predictive maintenance of the asset.