Combined Use of Modal Analysis and Machine Learning for Materials Classification.

Area of Science:

• Structural Dynamics and Material Science
• Computational Engineering and Machine Learning

Background:

• Modal analysis is crucial for structural dynamic testing of linear structures, ensuring material safety and preventing failures.
• Understanding modal parameters is key to assessing material integrity and predicting potential failure modes.

Purpose of the Study:

• To develop and validate a novel machine learning approach for classifying engineering materials based on modal analysis.
• To establish a classification method solely dependent on a material's resonance frequency.

Main Methods:

• Performed modal analysis using ANSYS to extract modal parameters, including mode number and associated frequency.
• Employed a machine learning approach to analyze the relationship between extracted modal variables for material classification.
• Validated the classification concept for both isotropic and orthotropic material types.

Main Results:

• Achieved 100% accuracy in classifying isotropic and orthotropic materials using the developed machine learning model.
• Demonstrated a strong correlation between mode number and resonance frequency for material identification.
• The new classification method relies exclusively on the resonance frequency of a material.

Conclusions:

• The study successfully presents a highly accurate, machine learning-based material classification method using modal analysis.
• Resonance frequency is identified as a key, sufficient parameter for distinguishing between different engineering materials.
• This research opens avenues for creating a single, versatile device for identifying and classifying various engineering materials.