Split spectrum algorithms rely on instantaneous phase information-a geometrical approach [US NDE].

Area of Science:

• Ultrasonic Nondestructive Evaluation
• Signal Processing
• Acoustics

Background:

• Split spectrum algorithms are effective for ultrasonic nondestructive evaluation (NDE) but are sensitive to parameter tuning.
• Current algorithms often lack a clear theoretical basis, relying on heuristic approaches without detailed signal and noise models.
• Understanding how phase and amplitude information is utilized remains a challenge.

Purpose of the Study:

• To initiate a model-based approach for split spectrum algorithms in ultrasonic NDE.
• To provide geometrical interpretations of the split spectrum concept.
• To evaluate conventional algorithms based on a new theoretical understanding.

Main Methods:

• Development of a detailed signal model for ultrasonic waves.
• Geometrical interpretation of the split spectrum technique to analyze phase and amplitude information.
• Evaluation of conventional algorithms (polarity thresholding, minimization, geometric mean) using the new model.
• Experimental verification of the signal model with real ultrasonic signals.

Main Results:

• A theoretical framework is established for understanding split spectrum algorithms.
• Geometrical interpretations reveal how algorithms utilize signal phase and amplitude.
• Conventional algorithms are critically evaluated based on the developed signal model.
• The proposed signal model is validated through experimental testing on ultrasonic data.

Conclusions:

• The model-based approach provides a foundation for optimizing split spectrum algorithms.
• Geometrical insights enhance the understanding of algorithm behavior in ultrasonic NDE.
• This work bridges the gap between heuristic algorithm design and model-driven signal processing.