Artificial Neural Network for Optimizing Gamma Radiation Shielding

Area of Science:

• Nuclear Engineering
• Computational Physics
• Machine Learning

Background:

• Gamma radiation shielding is crucial for medical, industrial, and research applications.
• Traditional Monte Carlo simulations for shield optimization are effective but computationally intensive and time-consuming.

Purpose of the Study:

• To investigate the efficacy of Artificial Neural Networks (ANNs) in identifying optimized gamma radiation shields.
• To develop a faster and more efficient method for shield design compared to conventional simulations.

Main Methods:

• Utilized MCNPX Monte Carlo code for initial simulations with a proposed shielding material.
• Generated a large dataset from simulations to train the ANN model.
• Validated ANN predictions against Monte Carlo simulation results and performed dose calculations using a water phantom.

Main Results:

• Achieved a prediction accuracy with less than 1% deviation between ANN and Monte Carlo simulations.
• Demonstrated the ANN's capability to accurately predict shielding performance for unknown configurations.
• Evaluated dose reduction using a water phantom to further optimize shielding material selection.

Conclusions:

• ANNs offer a powerful and efficient tool for radiation shielding studies.
• The developed ANN model accurately predicts optimal weight fraction combinations for shielding materials.
• This approach accelerates the design process for effective gamma radiation shields.