FIGUEROA, MARICELA (CINVESTAV - IPN), MARTINEZ-GARCIA, Juan Carlos (CINVESTAV-IPN), BONILLA, Moises (CINVESTAV - IPN), MALABRE, Michel (CNRS- UMR6597, (Ec. Centrale - Ec. de Mines - Univ. de Nantes)

Illustrating an Inverse-Based Failure Detection and Identification Technique Based on Reconstruction of Unknown-Inputs

Scheduled for presentation during the Regular Session "Linear system identification I" (MoP09), Monday, July 24, 2006, 17:00−17:25, Room J

17th International Symposium on Mathematical Theory of Networks and Systems, July 24-28, 2006, Kyoto, Japan

This information is tentative and subject to change. Compiled on April 26, 2024

Abstract

This paper illustrates a failure left-inverse-based detection and identification technique (for linear time-invariant systems) via a common example consisting of a mathematical model of a hydraulic Four-Tanks system affected by two kinds of possible failures: leakages in the Tanks and cloggings in the Pipes linking the Tanks. Both classes of failures are considered as unknown-inputs affecting the monitored system, and the detection mechanism is based on the left-inversion of the monitored system in order to reconstruct the unknown-inputs, which modelize both the leakages and the cloggings as fictitious failed actuators. We compare the inverse-based technique (which produces a derivative detector implemented via exponential approximation) with a classic observer-based detector (which performs residual generation via state identification).