Real-time Estimation of States & Unknown Boundary Conditions
Part of my PhD research at the University of Adelaide. We use the Elastic Water Column Model (EWCM) to build a linear, control-ready model of a pipeline, then combine a High-Gain Observer (HGO) with a Sliding-Mode Observer (SMO) to reconstruct both internal states and unmeasured demands/boundary heads in real time.
Why utilities care
- Estimate unmeasured demands and unknown heads from pressure sensors only.
- Real-time alternative to batch EM/WLS/ITA—no long replays.
- Feeds alarms, control set-points, and what-if testing.
What’s different
Classical UIOs need matching conditions rarely met in hydraulics. We generate auxiliary outputs via HGO to recover matching, then an SMO reconstructs states and unknown inputs with robust injection.
Laboratory validation
Experiment in the Robin Hydraulics Laboratory: two tanks at the ends of a copper pipe, two internal pressure sensors. The method tracks changing boundary heads with <10% error despite model uncertainty and noise.
