DIANA: Creación de un modelo de normalidad multi -parametral para detección de malfuncionamientos en equipos de plataforma naval
Fecha
2019-11-20Tipo de documento
conferenceObjectMateria/s Unesco
3310.04 Ingeniería de MantenimientoResumen
An artificial intelligence system needs to have a complete and reliable database that
facilitates the identification and diagnosis of a failure when the indications that characterize it are
activated: its sensorized parameters. For this, a Failure Modes, Effects and Criticality Analysis for
Artificial Intelligence (FMECA-AI) or its translation: "Analysis of Failure Modes, Effects and
Criticality for Artificial Intelligence" of the studied system is carried out. In this way, not only the list
of failures is available, also the effects it produces, the criticality of the consequences and how they
are measured, and then implanted in a predictive system based on artificial intelligence.
For this it is necessary to cross the effects produced by each failure with the indicators that facilitate
the monitored signals in order to connect the detection subsystem with the fault database. This allows
us to obtain a diagnosis of what is the failure or the possible failures that are occurring when
anomalous variation is detected in the sensorized data.
The procedure used is based on the existing one used to create a fault database called FMECA. This is
included in the so-called Reability Center Maintenance (RCM) "maintenance based on reliability",
due it is a successful systematic and efficient procedure verified in aeronautics, defense and in the
industry in general.