Thickness prediction in metal alloys using nuclear techniques and artificial neural network: Modelling

Area of Science:

• Nuclear Physics
• Materials Science
• Computational Modeling

Background:

• Metal alloys are crucial in aerospace, automotive, and biomedical fields, necessitating reliable identification methods.
• Gamma densitometry offers a non-invasive approach for material characterization using radiation sources and detectors.
• Accurate alloy identification is vital for quality control and performance in critical industries.

Purpose of the Study:

• To develop and validate a method for predicting the thickness of various metal alloys using gamma transmission.
• To assess the efficacy of artificial neural networks (ANN) in analyzing gamma densitometry data for alloy identification.
• To evaluate the reliability and accuracy of the proposed technique across different alloy types and thicknesses.

Main Methods:

• Utilized MCNP6 code to simulate measurement geometry and generate a dataset for ANN training.
• Employed gamma transmission with a ¹³⁷Cs radiation source and scintillator detector.
• Trained an artificial neural network to predict the thickness of aluminum, titanium, and steel alloys (2-50 mm).

Main Results:

• Achieved successful thickness prediction for all tested metal alloys, including aluminum, titanium, and carbon steels.
• Demonstrated high accuracy, with over 96% of predictions having a relative error below 5% using ¹³⁷Cs radiation.
• Confirmed the technique's reliability even for alloys with similar density values, such as various aluminum alloys.

Conclusions:

• Gamma densitometry combined with ANN provides an effective and accurate method for non-invasive metal alloy thickness prediction.
• The developed technique shows significant potential for industrial applications requiring precise material identification and quality control.
• The use of ¹³⁷Cs radiation offers a reliable energy source for differentiating and measuring alloy thicknesses with high accuracy.