SOUZA, A. G. M.; http://lattes.cnpq.br/6381993209967387; SOUZA, André Gomes Medeiros de.
Résumé:
In simulators for operator training, generating a training scenario demands a lot
of time and creativity. In the SIMULOP software environment, widely used in the
Brazilian electric sector, the interface with the tutor, for scenario building, is not
ergonomic, demanding high efforts, time and creativity, in their description. The
development process of these scenarios is general, not accounting for the individual
training needs of operators. In this project, we propose the automatic generation of
training scenarios, for the SIMULOP environment, based on the analysis of log files of
the operation, during which failures happened in the system, as well as the descriptive
report of human error – RDFH. This is done in order to simulate real environment
occurrences aiming to avoid the reoccurrence failures of human origin. This project was
developed based on a tool, built to support the tutor when interactively building
Scenarios, which was developed during an R & D project with the company CHESF at
LIHM at UFCG. It combines the information from the analysis of RDFH reports with
that from incident logs, recovered from the supervisory system SAGE, also adopted by
the electricity sector. This project is based on two ontologies which were built to
describe a training scenario. From the ontologies, it is produced a plain text file which
can be run by the OTS in the SAGE / SIMULOP system. Thus, based on these two
ontologies it became possible to generate training scenarios from real events, which can
be stored in a training scenario database, which allows for the classification of events
and scenarios according to a typology of the human error.