Contribution to fault detection for dynamic processes using non-linear interval models

• Autores: Stelian Stancu Alexandru
• Directores de la Tesis: Joseba Jokin Quevedo Casin (dir. tes.), Vicenç Puig (codir. tes.)
• Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2004
• Idioma: español
• Tribunal Calificador de la Tesis: Teresa Escobet (presid.), Jordi Saludes (secret.), J. Patton Ron (voc.), Luc Jaulin (voc.), Jose Sa da Costa (voc.)
• Texto completo no disponible (Saber más ...)
• Resumen

  • Model-based fault detection is based on the use of mathematical models of the monitored system. However, modelling errors and disturbances in complex engineering systems are inevitable. Therefore, there is a need to develop robust fault detection algorithms. The robustness of a fault detection system means that it must be only sensitive to faults, even in the presence of model-reality differences. One of the approaches to robustness, known as active, is based on generating residuals which are insensitive to uncertainty, while at the same time sensitive to faults. This approach has been extensively developed these last years by several researchers using different techniques: unknown input observers, robust parity equations, H, etc.. But, in case of models with uncertainty located in the parameters, known as interval models, perfect decoupling of the residuals from uncertainties is only possible in a limited number of model parameters. In the case of unlimited number of uncertain parameters, there is a second approach, called passive, that enhances the robustness of the fault detection system at the decision-making stage, mainly propagating the effect of the parameter uncertainty to the residual that can be used as an adaptive threshold.

    Actually, several research groups are following this approach, also known, as the bounding or set-membership approach, because of the use of bounds to describe parameter and residual uncertainty. Set-membership approaches were first considered in the late sixties in the context of state estimation.

    More recently has received a renewed interest in connection with model identification for robust control and signal processing. Actually, there are several research groups in fault detection following this approach.

    The problem of robustness in fault detection using observers has been treated basically using the active approach, based on decoupling the effects of the uncertainty from the effects of the faults o