Radionuclide identification system using convolution neural network for environmental radiation monitoring

• DOI: 10.11591/ijece.v15i2.pp2282-2290
• Publication Date: 2025-01-26T14:07:36Z
• Authors: Istofa Istofa; Gina Kusuma; Firliyani Rahmati...; Joko Triyanto; I Putu Susila; Prawito Prajitno
• Keywords: Convolution (computer science); Identification; Radiation monitoring

Radionuclide identification is an important task for nuclear safety and security aspects, especially to environmental radiation monitoring systems. This study aims build automatic radionuclide system that can be applied in stations. The gamma energy spectrum was obtained by varying types, measurement time source distance using a scintillation detector. dataset collected converting into images, data pre-processing removing background noise normalizing the spectrum. Automatic demonstrated as development method based on convolutional neural network (CNN) algorithm, where images come from gamma-ray form of photoelectric peak characteristic. Three CNN architectures are used train model, which VGG-16, AlexNet Xception. performance each model evaluated accuracy, precision recall find appropriate architecture. most optimum results shown VGG-16 with accuracy 97.72%, 97.75% 97.71%. models critically reviewed it concluded developed further implemented embedded devices utilizing tiny machine learning (TinyML) platform