The MLAI Meetup is a community for AI researchers and professionals which hosts monthly talks on exciting research. Our format is:
- 6:00 - 6:20: Socializing
- 6:20 - 6:40: Announcements and AI news
- 6:40 - 7:40: Talk(s) and Q&A
- 7:40 - 8:00 Networking
- 8:00: Head to the nearest pub for dinner
Joanna Wieclaw: "From Fixed Thresholds to Learned Boundaries: Predictive Maintenance at the Australian Synchrotron"
Talk Description: Last year I showed you how the Australian Synchrotron works. This year I'd like your help keeping it working.
Much of my job in Controls involves troubleshooting equipment after something has already gone wrong. I'd rather catch the warning signs first.
I'm developing a proposal for a phased predictive-maintenance system using data we already collect through EPICS (Experimental Physics and Industrial Control System). The proposed first phase is deliberately simple: choose one system, set fixed upper and lower thresholds, and alert on breaches. In parallel, I'd like to explore what can be learned from our historical data using offline ML models, with signals such as encoder drift, motor current and vacuum levels among the first targets.
Later phases could augment those thresholds with learned boundaries and look for patterns across equipment rather than within a single system.
I'd like to hear how you would approach it: what methods are worth trying early, how you would validate models when labelled failures are scarce, and what I should be implementing now to make the data useful as the system develops.
Speaker Bio: Joanna Wieclaw is a Senior Controls Technician at the Australian Synchrotron (ANSTO). With a mechatronics background and extensive experience in system integration and live fault-finding, she spends much of her time diagnosing complex equipment under pressure.
