Short name: PreCoMInd
Long name: Predictive Cognitive Maintenance of Industry 4.0 systems
Call Stage 1: F4Fp-SME-1 (see call details)
Proposal number: F4Fp-SME-COD200721-01
SUMMARY REMARKS & TESTBEDS
Predictive Cognitive Maintenance of Industry 4.0 systems (PreCoMInd) addresses the challenge of operational expenditure reduction in large-scale Industry 4.0 systems containing cyber-physical as well as RAN devices in IoT.
PreCoMInd aims to predict the problems/failures during operation of the system using the semantic dataset (SD) integrating heterogeneous data. In the first phase, data about power consumption (PCD), spectrum sensing (SSD) and error logs (ELD) will be collected, semantically annotated and integrated into the SD. In the next phase, software tools would be developed to enable the integration of user-defined data types into the SD together with the model that would be used in reasoning as well as machine learning-based AI algorithms for prediction and anomaly detection.
State-of-the-art solutions for predictive maintenance are not self-explanatory and consequently can not be further improved by contextualization with error logs. PreCoMIndproposes to integrate semantic annotations of the system’s data with the background domain knowledge. The objective of the first phase is to develop a prototype to answer the two questions crucial for the product commercialization: 1) are the three proposed data sources enough to deliver commercially viable value-add, and 2) what combination of AI tools would deliver the best results in the production environment.
- PreCoMInd experiment (download the poster)
- Review PreCoMInd experiment | FEC8 (download the slides)