In May 2019, a consortium of six European institutions started developing a new approach for anomaly detection based Industrial Internet of Things (IoT). The fundamental idea was to develop a way to articulate in a more organic manner the steps of data acquisition, transmission, fusion, and analytics to identify hard to find “informative outliers”. The proposed idea is called Framework for the Identification of Rare Events via MAchine learning and IoT Networks (FIREMAN), which has been developed to be general enough to be deployed in different context with flexibility to incorporate a diverse set of requirements. At this point, we have produced several journal publications, master theses, and open-source codes; all reported in deliverables.
At this point, the proposed solution has been deployed and tested in different contexts and testbeds, including the plant of the SEAT car manufacture in Spain, a microgrid testbed in National Technical University of Athens via ERIGRID 2.0, the Smart Campus from Oulu University, and a power electronics testbed in Aalborg University. Currently, the proposed solution is being further developed to move towards higher Technology Readiness Levels (TRLs) to incorporate physics-based machine learning by leveraging simulation models as part of data pre-processing at the acquisition step combined with imputation methods based on their specific end use, building upon a semantic-functional communication system (a new concept to be developed beyond 5G IoT networks). In this way, the idea is to combine FIREMAN fundamental research with the expertise in intelligent machines, physics-based simulations, and business models to realize its full potential.
More information: https://fireman-project.eu/
Author: Assoc. Prof. Pedro Nardelli, FIREMAN coordinator and MORE SIM Group Leader (IoT solutions)