Predicting alpha and gamma indexes from industrial recyclates using artificial intelligence

Oyebisi, Solomon and Khalid, Al kaaf and Shammas, Mahaad I and Seyam, Mohammed and Oyewola, Miracle Olarenwaju (2025) Predicting alpha and gamma indexes from industrial recyclates using artificial intelligence. Predicting alpha and gamma indexes from industrial recyclates using artificial intelligence, 7 (1370). pp. 1-27. ISSN 3004-9261

Text - Published Version Download (3MB)

Abstract

Industrial recyclates are used in the building and construction sector owing to the environmental impact of cement manufacturing. However, radiation exposure from
these alternative materials could endanger the environment and human health. In light of this, this study assessed the radiological characteristics of recycled industrial
waste materials by paying particular emphasis to their activity concentrations and evaluating their radiation levels. Deep neural networks of different network structures were applied to model the input arguments and target data. The model’s performance was assessed and validated. Radiation hazards are shown to be present in the majority of these industrial byproducts. Compared to other network structures, 1-4-4-4-1 and 3-14-14-14-1 architectures gave the best performance metrics for alpha and gamma indexes. The validation of untrained data with the developed model exhibited a strong relationship with 0.9997 and 0.9708 R2 for the alpha index and gamma index. Thus, the deep neural network is an ideal option for predicting the radiation from industrial byproducts.

Item Type:	Journal Article
Uncontrolled Keywords:	Activity concentrations, Artificial intelligence, Hazards, Modelling, Radiation, Recycling
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) T Technology > TH Building construction
Divisions:	School of Information Technology, Engineering, Mathematics and Physics (STEMP)
Depositing User:	Solomon Oyebisi
Date Deposited:	08 Dec 2025 03:46
Last Modified:	08 Dec 2025 03:46
URI:	https://repository.usp.ac.fj/id/eprint/15202

Actions (login required)

View Item