Reproduce “PMI-DT: Leveraging Digital Twins and Machine Learning for Predictive Modeling and Inspection in Manufacturing” on OpenResearch

About this paper

Chas Hamel, Md Manjurul Ahsan, Shivakumar Raman

A Predictive Maintenance and Inspection Digital Twin (PMI-DT) framework integrates real-time dimensional inspection data from a CyberGage 360 system with machine learning models to predict component failures in manufacturing. The framework accurately predicted bolt failures with 100% classification scores, identifying maximum position and maximum load as critical indicators for Nylon-12 material.

Key insights

• The PMI-DT framework effectively integrates physical twin fabrication, advanced vision inspection (CyberGage 360), and machine learning to create a robust predictive quality assurance system.
• Maximum Position and Maximum Load were identified as the most significant features for predicting bolt failure, aligning with the brittle material properties of Nylon-12 bolts under tensile stress.
• The study validates the feasibility of using Digital Twins combined with real-time data streams and predictive machine learning models to transition from reactive to proactive quality control and maintenance in manufacturing.

Summary by alphaXiv.