Loading...
Publication
A multi-agent architecture for diagnosing simultaneous faults along water canals
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 2.02 MB | Adobe PDF |
Advisor(s)
Abstract(s)
Water is intensively used in mankind activities, in particular in agriculture. Water is commonly conveyed
for agriculture purposes through water canal networks which are large-scale spatially distributed
systems crossing extensive regions. In the presence of leaks, unauthorized water withdrawals, water
depth sensor faults or gate faults, the quality of service can be severely compromised. A system able
to diagnose which type of fault is present at a given time is of vital importance to access the current
state of the water canal and proceed to restore its nominal condition. This paper proposes a multiagent
architecture to simultaneously detect, isolate and estimate lateral outflows (e.g., leaks or water
withdrawals) and hardware faults (e.g., a gate obstruction or a downstream water depth sensor fault) in
water canal networks. First, the main canal network is broken down into several subsystems composed
of a single canal pool with the corresponding gate. Then, an agent is assigned to each subsystem aiming
at its fault diagnosis. The approach is based on the generation and evaluation of residuals obtained from
the comparison of model-based output signals with real data. Application to an experimental water canal
bears out the proposed architecture as a valuable tool for monitoring and supervising general water
canals.
Description
Keywords
Citation
Nabais, J., Mendonça, L. F. & Ayala Boto, M. (2014). A Multi-Agent Architecture for Diagnosing Simultaneous Faults along Water Canals. Control Engineering Practice, 31, pp. 92-106