AI-Powered Corrosion Detection: A Game-Changer for Industrial Maintenance

Researchers at the Indian Institute of Science (IISc) and the Qatar Science and Technology Research Center (QSRTC) have developed a cutting-edge automated method to assess corrosion in industrial equipment using advanced machine learning and image analysis. This novel technique, published in npj Materials Degradation, can analyze microscope images of corroded metal surfaces to estimate corrosion severity without human input. The AI-based approach has far-reaching implications for industries like power generation and oil/gas, where corrosion poses significant economic and safety challenges.

Key Takeaways:

• The researchers developed a machine learning algorithm that can analyze microscope images of corroded metal surfaces to estimate corrosion severity without human input.
• The algorithm focuses on two critical indicators of corrosion: the thickness of corrosive deposits on the surface of metals and the porosity (number of tiny holes) within those deposits.
• The algorithm can quantify these characteristics and infer key features that indicate how much corrosion has happened, including the concentration of corrosive chemicals and the acidity of the environment beneath the deposits.
• The researchers tested their method in checking the under-deposit corrosion (UDC) of steam generator tubes, which is a particularly challenging and prevalent form of corrosion in industrial boilers.
• The algorithm is accurate, getting it right about 73% of the time, and is faster and more consistent than having people manually examine the optical microscopy images to determine the severity of corrosion.
• The technique employs k-means clustering, a machine-learning technique that segments microscopy images into distinct regions without prior assumptions about their appearance.
• The approach can be adapted to various corrosion product morphologies but needs to be tailored to each specific case, as morphological features differ across corrosion mechanisms.
• The researchers caution that the next critical step is to validate the algorithm on much larger and more diverse datasets.

Statistics:

• The algorithm achieved an accuracy of 73% in detecting corrosion in steam generator tubes.
• The algorithm can infer key features that indicate how much corrosion has happened, including the concentration of corrosive chemicals and the acidity of the environment beneath the deposits.
• The researchers identified specific pH levels that indicate when corrosion is worsening, including a threshold at pH ~2.8-3 that delineates the transition from Stage 3 to Stage 4.

Sources:

• npj Materials Degradation
• Indian Institute of Science (IISc)
• Qatar Science and Technology Research Center (QSRTC)