This work is devoted to activity recognition in the setting of data analysis in aeronautics. Formal methods are applied to the cer-tification and safety analysis processes of Unmanned Aircraft Systems in breakdown situations. The behaviour of these systems in case of a failure is entirely modeled and implemented. A temporal language — the Chronicle language — describes arrangements of events which are employed to detail undesired circumstances that would lead to breaches in safety. A C++ chronicle recognition tool is used to recognise all the possible occurrences of these situations as soon as they occur.