COMET project inFADIA (2015-2018)
Industrial Fault Detection, Diagnosis and Prognosis
Aims / Research Topics
The inFADIA project focusses on the detection and isolation (e.g. location and classification) of fault events in large industrial environments. Major challenges are the restrictive sensor setting and naturally contaminated data. The project develops robust methods for fault detection and isolation which cope with these problems. Based on production data of a single device or multiple similar devices, faults are detected, located in time and the possible type (respectively a possible cause) of the fault event is determined. In the course of the project various use cases are addressed, e.g. detecting and diagnosing faults in photovoltaic systems, injection molding machines and mobile stone crushers.
Methods / Software / Proof of Concepts
- Algorithm framework for fault detection and isolation
- Fault message board: Web application for reporting and visualizing detected and isolated faults and metadata for a large amount of systems. Faults are ranked and sorted by their relevance
- Fault web interface: REST API for querying information about detected and isolated faults
- Algorithm for a virtual balance as the basis for detection of faulty operations
- Infrastructure for on-machine deployment of fault and wear detection algorithms
- Locating faults in photovoltaic systems data [KT17a]
- Robust learning of neighborhood relationships to identify comparable machines and devices [TGG15]
- Fault identification and generation of meaningful alarms from fault detection [KT17c]
This project is subsidized in the frame of COMET – Competence Centers for Excellent Technologies by BMVIT, BMDW, State of Upper Austria and its scientific partners. The COMET program is handeled by FFG.