The Total Instantaneous Blockage (TIB) of an assembly in the core of a Sodium-cooled Fast Reactor (SFR) is under study. The dataset is provided by the Atomic Energy and Alternative Energies Commission (CEA). Through the estimation of the temperatures on the top of each assembly of the SFR's core, an original strategy for the detection of a TIB is proposed and implemented to a hypothetical simulated incident. Since there is no knowledge of the system's model, it is considered as a blackbox and the estimation of temperatures is made by a Neural Network.