PhD in Computing · Curtin University · 2026 – now
Predictive maintenance that tunes itself.
My research asks whether an explicit heuristic — built on sampling frequency, warning horizon and signal characteristics, plus a contextual operational memory — can configure MiniRocket, XGBoost and lightweight LSTM models on edge devices across heterogeneous mining assets, without manual expert tuning.
Proposal · Self-Adjusting Hyper-parameters for MiniRocket, XGBoost and Lightweight LSTM in Edge-Based Predictive Maintenance for Mining
Interactive · simulated dataThe predictive-maintenance twinInject a bearing fault on a mining conveyor drive and watch the model catch it: vibration spectrum, health index, remaining useful life and the maintenance call.Open the twin →Knowledge universe
Six ideas, one research question.
Every galaxy is a topic found in the proposal by c-TF-IDF; every star is one of its top concepts. Lines join concepts that several topics share. Drag to travel, tap a galaxy to visit it.
The full research dossier
The proposal's architecture, evidence pipeline (PRISMA), lexical and semantic analysis, quality audits and the immersive research orbits — the complete working view.