Predictive Maintenance (PdM)

Practical example and current challenges

• In the manufacturing sector, the monitoring of industrial plant is crucial in order to ensure effective and sustainable operation. The maintenance that has to be undertaken can either be reactive or preventative. The choice is between accepting that machines will break down and production will come to a standstill, or to invest in new spareparts before they are needed and thus creating unnecessary costs.
• By detecting anomalies within data streams and processes, critical situations can be prevented before they occur. Artificial intelligence (AI) and machine learning (ML) can be used to identify or even predict malfunctions. This enables action to be taken early on.
• However, when it comes to predictive maintenance and system optimisation, the problem is often insufficient stock of data or no data at all, and this is needed in order to use AI applications effectively.

What added value does the "GAIA-X project" offer?

• GAIA-X creates an overarching infrastructure that helps generate a generic data model from different data and enables the probabilities of plant failure to be calculated. The availability of data cross-company can increase the breadth of data available and improve the accuracy of predictions.
• Also, GAIA-X provides secure and autonomous data storage through a particular authorization management system whereby data for training ML models is stored separately from dedicated "decision data" for maintenance at the production plants.
• Furthermore, the GAIA-X infrastructures ensure that the ML algorithms are continuously trained, and the AI models are stored securely while ensuring data autonomy and control are maintained.

Use Case Team

• Olga Mordvinova – incontext.technology
• Dr. Andrea Rösinger – FORCAM