Combining interval and qualitative reasoning for fault diagnosis
| Authors: | Calderón-Espinoza Gabriela, Universitat de Girona, Spain
Armengol Joaquim, Universitat de Girona, Spain
Sainz Miguel Á., Universitat de Girona, Spain
Herrero Pau, Universitat de Girona, Spain |
|---|
| Topic: | 6.4 Safeprocess |
|---|
| Session: | Robust Fault Detection and Isolation |
|---|
| Keywords: | Fault detection, Fault diagnosis, Intervals, Uncertainty, Uncertain Dynamic System |
|---|
Abstract
A methodology is proposed to diagnose faults from a detection system that allows to express uncertainty with intervals in parameters and measurements. A fault is detected by comparing current measurements with computed external and internal approximations of behavior. The results are applied to diagnose faults. A qualitative data set is obtained in order to create fault rules. A fault library is built from these rules to make fault diagnosis. Diagnosis is given by a set of the most possible faults with similar behavior.
