Tire mode shape categorization using Zernike annular moment and machine learning classification.

Area of Science:

• Mechanical Engineering
• Computational Science

Background:

• Tire mode shape categorization is crucial for optimizing driving performance and safety.
• Current methods for mode categorization are labor-intensive and require significant manual interpretation.
• Developing accurate digital twins for tire analysis necessitates efficient mode shape classification.

Purpose of the Study:

• To establish a database of categorized tire mode shapes using identified features.
• To develop a machine learning (ML)-based surrogate model for automated tire mode shape classification.
• To eliminate the need for manual effort in interpreting tire modes.

Main Methods:

• Utilized the Zernike annular moment descriptor (ZAMD) to create feature maps of tire mode shapes.
• Employed Modal Assurance Criteria (MAC) correlation values with ZAMD for mode categorization and labeling.
• Implemented supervised learning algorithms including decision trees, random forests, and XGBoost for surrogate model development.

Main Results:

• Successfully created a database of categorized tire mode shapes.
• Developed an ML-based surrogate model capable of classifying tire mode shapes.
• Achieved a high classification accuracy of 99.5% with the best-performing model, eliminating manual effort.

Conclusions:

• The proposed ML framework effectively automates the categorization of radial tire mode shapes.
• This automated approach significantly enhances the development of digital twins for tire performance analysis.
• The study demonstrates a highly accurate and efficient method for tire mode shape classification, improving safety and performance prediction.