Behaviour simulation in Consistency-based Diagnosis requires knowing the initial value. This assumption is not easily fulfilled in real systems, even in the presence of measurements related to state variables due to noise and parameter uncertainties.

This work proposes the integration of state observers to estimate initial states for simulation in consistency-based diagnosis with possible conflicts, using the BRIDGE framework, proposing an extension for a class of dynamic systems. Suitable state-observer structural models are obtained through same algorithms used to find possible conflicts – minimal subsystems with analytical redundancy–, without additional knowledge in the models.