Interpolation techniques for ultrasonic data.

Area of Science:

• Acoustics and Ultrasonics
• Computational Mechanics
• Data Science

Background:

• Limited experimental ultrasonic data hinders the full exploitation of diagnostic applications.
• Generating sufficient data via finite element (FE) modeling is computationally expensive.
• This data scarcity impacts non-destructive evaluation (NDE) method qualification and machine learning algorithm training.

Purpose of the Study:

• To develop a computationally efficient method for generating large-scale ultrasonic datasets.
• To augment scarce experimental ultrasonic data using interpolation techniques combined with FE modeling.
• To validate the accuracy and applicability of the proposed data generation method.

Main Methods:

• Utilized interpolation techniques in conjunction with pre-existing finite element (FE) model results.
• Developed and presented methods for rapid ultrasonic data generation without solving additional FE models.
• Validated the approach through four exemplary cases of increasing complexity.

Main Results:

• The interpolation method successfully generated ultrasonic data comparable to full FE modeling.
• Results from the proposed method were typically within 1% of expected values.
• Data generation was achieved at a fraction of the computational cost of traditional FE modeling.

Conclusions:

• The proposed method offers a computationally efficient solution for generating realistic ultrasonic data.
• This technique can significantly enhance data availability for machine learning and NDE applications.
• The method is validated for various physical setups and signal complexities, with broad applicability.