This article focuses on the professional training of operators of Nuclear Power Plants (NPP) on Full-Scope Simulators (FSS). In such context, observation and analysis of the individual and collective interaction of the trainees are critical and particularly difficult. The objectif of our work is to propose models and tools to help trainers observe and analyze trainees' activities during preparation and debriefing. For that purpose, our approach consists in representing the actions of the operators and the simulation data in the form of modelled traces. These modelled traces are then transformed in order to extract higher information level. Trainers can visualized the different levels of trace to analyze the reasons, collective or individual, of successes or failure of trainees during the simulation. In order to validate our approach, we have developed the prototype D3KODE based on the trace model and transformation that we proposed. This prototype was then evaluate according to a protocol based on a comparative method in the context of several experiment conducted with a team of experts, trainers and trainees from EDF Group.