The work presented in this paper deals with the diagnosis of single and multiple faults for continuous dynamic systems. It consists on developing a global diagnosis strategy for the operating modes management in both normal and abnormal situations. We fi rst developed a new graphical formalism for dynamic system modelling. This formalism is emanating from the BG methodology and it is called Signed Bond Graph (SBG). This latter is easily understandable by a number of properties and defi nitions that we have established. The development of such formalism allows to use structural and causal properties of the BG and to expand its scope to include qualitative reasoning. Furthermore, we proposed a generic model for integrating functional Generic Component Models (GCM) and SBG models for the management of operating modes and reconfi guration conditions of an autonomous system using a finite automaton. Finally, we proposed a method for diagnosing both single and multiple faults using an abduction approach based on the faults propagation within the SBG by starting from a set of observations. The proposed methodology is validated by two di erent systems namely a proton exchange membrane fuel cell and an electromechanical system of an electric vehicle.
