The mixed logical dynamical (MLD) formalism has proved to be an efficient modeling framework for hybrid systems described by dynamics, logic and constraints. Furthermore, it allows formulating and solving practical problems such as moving horizon control and state estimation. Starting from the general framework of state estimation, this paper presents a modified version adapted to fault detection of hybrid systems under the MLD formalism. A decomposition into several sub-problems decreases the computation time for a faster real time implementation. This strategy is applied in simulation to fault detection of sensors of a steam generator benchmark.