A model for estimating the phase fraction in the microstructure of densified graphite iron (CGI) using machine...

Area of Science:

• Materials Science
• Metallurgy
• Computer Science

Background:

• Alloy properties are assessed via microstructure image analysis, influenced by alloying elements and heat treatment.
• Manual microstructure phase assessment is time-consuming, requires expertise, and can be inconsistent.
• Accurate microstructure analysis is vital for controlling production processes and preventing defects in cast iron.

Purpose of the Study:

• To develop a machine learning algorithm for automated prediction of matrix phases (pearlite, ferrite, graphite) in vermicular cast iron microstructures.
• To accelerate and enhance the accuracy of metal melt inspection through automated image analysis.

Main Methods:

• Utilized machine learning techniques to analyze microstructure images of vermicular cast iron.
• Developed an algorithm to predict the presence and distribution of key matrix phases.

Main Results:

• Successfully created a machine learning algorithm capable of identifying pearlite, ferrite, and graphite in vermicular cast iron microstructures.
• The developed method offers a pathway to automate the metal inspection process.

Conclusions:

• Machine learning presents an effective solution for automating microstructure phase analysis in vermicular cast iron.
• Automated analysis can significantly improve the efficiency and reliability of metal melt inspection.