Anomaly detection is an important issue heavily investigated within different research areas and application domains. Its application in the industrial systems sector may be essential also for the protection of critical infrastructures. Due to the huge amount of involved data and to their complexity the use of machine learning may be the clue. The basic idea is describing an industrial process by a series of key attributes whose measures (the features) compose a state vector including heterogeneous types of measurements. Each feature should be a key attribute which can help discriminate between a normal functioning condition and an anomaly. In this context, the paper presents the use of a deep neural network architecture called autoencoder to detect anomalies due to either system faults or cyberattacks. The chosen application field is a photovoltaic system connected to the grid. The results, even if preliminary, are really promising.

Neural network architecture to detect system faults / cyberattacks anomalies within a photovoltaic system connected to the grid

Gaggero G. B.;Rossi M.;Girdinio P.;Marchese M.
2019-01-01

Abstract

Anomaly detection is an important issue heavily investigated within different research areas and application domains. Its application in the industrial systems sector may be essential also for the protection of critical infrastructures. Due to the huge amount of involved data and to their complexity the use of machine learning may be the clue. The basic idea is describing an industrial process by a series of key attributes whose measures (the features) compose a state vector including heterogeneous types of measurements. Each feature should be a key attribute which can help discriminate between a normal functioning condition and an anomaly. In this context, the paper presents the use of a deep neural network architecture called autoencoder to detect anomalies due to either system faults or cyberattacks. The chosen application field is a photovoltaic system connected to the grid. The results, even if preliminary, are really promising.
2019
978-1-7281-3729-2
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11567/1017169
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 2
social impact