FDI based on causal graphical approaches for nonlinear processes

In this paper, fault detection and isolation for non linear systems using causal graphical approaches are presented. A diagnosis based on signed directed graph (qualitative model) is provided by determining initial and final responses using forward and backward propagation. Moreover, a bond graph (quantitative model) is used to generate Analytical Redundancy Relations (ARRs) through causal paths. To illustrate the effectiveness of the proposed approaches, a thermofluid system is studied and some simulations are provided.