COMET-Project FDI (2019-2022)

Machine Learning Based Fault Detection and Identification

Motivation

According to current studies machine learning will have or already has a very large impact on predictive maintenance. Within the scope of this project, robust data-driven and model-based algorithms for machine learning will be developed for fault detection, identification and prediction. We will evaluate specialized approaches such as causal discovery (see also the COMET project SmartDD) and robust machine learning for available data. Potentially useful data sources are sensors and system logs that contain alarm and fault messages. In combination with stream and big data processing solutions, continuous monitoring and analysis of possible large quantities of devices, machines and industrial plants will be enabled and thereby improve maintenance and production. Examples of devices, machines and facilities include hydraulic systems, machinery with rotating elements and plastics machinery.

Expected Results

  • Efficient algorithms will provide support for robust learning, outlier detection, robust stochastic optimization, and event time prediction
  • These algorithms are the basis for software tools for predictive and optimal maintenance tailored to the needs of the company partners
  • In addition to algorithm implementations, our software will handle stream processing, big data processing, and visualizations

Funding Partner

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 handled by FFG.

Contact

Michael Roßbory

Roßbory Michael

Researcher Data Analysis Systems
Phone: +43 50 343 860

back