Ce cours est une sensibilisation à des techniques de la science des défaillances. Il illustre par de nombreuses méthodes du diagnostic des défauts, en précisant ce que l'on doit diagnostiquer et comment on le fait. Les aspects homme-machine y sont traités. Le diagnostic, suscite depuis les années 1970 un intérêt croissant tant au niveau du monde industriel que de la recherche scientifique, dans l’objectif vis à développer des systèmes de diagnostic en vue d'assurer la maintenance intelligente ou prédictive des industriels, donc l'objectif est de développer des générateurs de résidus robuste (indicateurs de défauts) de l'installation qui sont utilisés pour détecter le dysfonctionnement et le localiser.
Abstract :
This course is concerned with the diagnosis of dynamical systems based on observers and their application to linear dynamical systems. This type of system is found in particular in the conduct of industrial chemical processes and can be considered as a prototype of many industrial applications in the process industry, such as chemical and petrochemical plants, oil and gas systems. The objective is to provide a robust residual generator for the detection of actuator faults or observer-based sensor faults. After the presentation of a state of the art on model-based diagnostic techniques. this course proposes, a strategy of estimation of the continuous states of a linear system subjected to the unknown input whose observer based on the perfect modeling of the systems in the presence of the matrices of the unknown input and the defect matrices is developed. On the other hand, two diagnostic procedures are presented.The first a proportional observer in the second chapter and the second with unknown input with perfect decoupling for robust residual generation in the third chapter to detect and isolate faults. In the last chapter we have exposed the theoretical foundations of the observer-based diagnostic technique with unknown inputs with perfect decoupling in order to estimate actuator and sensor faults for fault isolation and to avoid false alarms. The two diagnostic procedures are applied for the detection of actuator faults and sensor faults. The application of the latter strategy will allow the robust generation of residuals at the defects of the system studied, in order to ensure the generation of the robust residuals and the performance of the system to be monitored.
Keywords: Unknown inputs, Diagnosis, Observer based diagnosis, Observer with unknown inputs.
- Enseignant: Souad TAHRAOUI