Abstract
An algorithm for active fault detection and isolation is proposed. In order to observe the failure hidden due to the normal operation of the controllers or the systems, an optimization problem based on minimization of test signal is used. The optimization based method imposes the normal and faulty models predicted outputs such that their discrepancies are observable by passive fault diagnosis technique. Isolation of different faults is done by implementation a bank of Extended Kalman Filter (EKF) where the convergence criterion for EKF is confirmed by Genetic Algorithm (GA). The method is applied to the two tank bench mark example in presence of two faults.
| Original language | English |
|---|---|
| Title of host publication | ICSE 2009 Twentieth International Conference on Systems Engineering |
| Number of pages | 6 |
| Publisher | IEEE Press |
| Publication date | 2009 |
| ISBN (Print) | 978-1-4244-3452-7 |
| Publication status | Published - 2009 |
| Event | ICSE 2009 Twentieth International Conference on Systems Engineering - Coventry, United Kingdom Duration: 8 Sept 2009 → 10 Sept 2009 Conference number: 20 |
Conference
| Conference | ICSE 2009 Twentieth International Conference on Systems Engineering |
|---|---|
| Number | 20 |
| Country/Territory | United Kingdom |
| City | Coventry |
| Period | 08/09/2009 → 10/09/2009 |
Keywords
- Active fault detection, hybrid systems, optimization, mixed logical dynamical system, mixed integer quadratic programming, EKF, GA.