Detecção e isolamento de falhas no sistema de atuação de foguetes utilizando grafos de ligação (bond graph)

Abstract

As new studies and new technologies have developed, modern systems have become more complex, but such equipment is still susceptible to failure. When these faults are previously identified, the maintenance process becomes simpler and cheaper. Some fault detection and identification techniques are known and proposed in order to determine their type and location through parameter estimates or methods based on real system modeling. Model-based techniques impose robustness on the system with rapid detection and isolation. This dissertation aims to study the fault detection and isolation (FDI) in the bond graph modeling rocket actuation system (Bond Graph, BG) in the 20-sim software. Binding Graphs have a very intuitive approach as they are based on the principle of energy conservation and the graphical representation facilitates the development of the FDI method. The fault identification process consists of two steps: the first step establishes the constituent equations from BG for each junction connected to a sensor, while the second step deals with the elaboration of a fault matrix that allows checking each fault signal and according to the signature isolate it. For this dissertation the electrohydraulic servo valve will be presented in a BG modeling structure that facilitates the derivation of analytical models, eliminating unknown variables and generating Redundant Analytical Relations (ARR) which are derivative constraints that provide the data for the failure matrix.