Artificial Intelligence for Corrosion Science

Involved People: Matteo Olgiati, Kai Schwenzfeier

Metals degradation due to (electrochemical) corrosion is ubiquitous in many engineering fields (aerospace, automotive, infrastructure, etc.) and prediction, prevention and maintenance are costly despite necessary to avoid major failures. On the other hand, understanding corrosion processes from the very early stages of initiation is sometimes demanding due to the multiple environmental and microstructural conditions that could determine both the thermodynamics and kinetics.

In our group, we aim to understand corrosion mechanisms of engineering alloys (e.g., aluminium alloys, steels, etc.) with a novel real-time and multi-sensorial approach, involving different detection and characterization techniques (electrochemical characterization, acoustic emission, optical microscopy and ICP-MS). Furthermore, we apply machine-learning algorithms to identify and recognize specific and/or recurrent patterns in the data collected, which will eventually lead to more reliable and damage-tolerant predictions on the overall state of corrosion.