An artificial neural network (ANN) code is carried out to estimate the radiological concentration of environmental samples, clay, and sand. The ANN modeling has been calibrated and validated with the measuring data acquired from an HPGe detector. Radioactivity was determined using a gamma spectrometric detector (HPGe) for the collection of clay and sand samples from Al-Tod, Luxor, Egypt. The radiological levels of the natural radioactive series of Ra, Th, and K-isotope have been estimated for the sand and clay samples. The average specific activity of Ra, Th, and K for clay-samples were 14.43 ± 0.5 Bq/kg, 15.8 ± 0.6 Bq/kg, and 273.2 ± 6.1 Bq/kg, respectively, where it for sand-samples were 12.4 ± 0.46 Bq/kg, 13.4 ± 0.5 Bq/kg, and 215.4 ± 15.11 Bq/kg, respectively. The feasibility of the ANN model has been studied by comparing its outcome data with the measured one. However, a robust correlation is observed for the radiological levels achieved by the ANN model and for the measured one in an error of around 5 percent. Therefore, besides measurement techniques, the ANN model could be a promising candidate for estimating and predicting the radiological levels from environmental samples. However, the use of the ANN model will reduce the measurement time and cost over many samples.

中文翻译：

---

利用人工神经网络模型估算环境样品放射性水平的可行性研究

执行人工神经网络 (ANN) 代码来估计环境样本、粘土和沙子的放射性浓度。 ANN 模型已使用从 HPGe 探测器获取的测量数据进行了校准和验证。使用伽玛能谱检测器 (HPGe) 测定从埃及卢克索 Al-Tod 收集的粘土和沙子样品的放射性。对沙子和粘土样品的 Ra、Th 和 K 同位素天然放射性系列的放射性水平进行了估计。粘土样品的 Ra、Th 和 K 的平均比活度分别为 14.43 ± 0.5 Bq/kg、15.8 ± 0.6 Bq/kg 和 273.2 ± 6.1 Bq/kg，而砂土样品的平均比活度为 12.4 ± 0.46分别为 Bq/kg、13.4 ± 0.5 Bq/kg 和 215.4 ± 15.11 Bq/kg。通过将其结果数据与测量数据进行比较，研究了 ANN 模型的可行性。然而，我们观察到 ANN 模型获得的放射水平与测量的放射水平存在很强的相关性，误差约为 5%。因此，除了测量技术之外，人工神经网络模型可能是估计和预测环境样本放射性水平的有前途的候选者。然而，使用 ANN 模型将减少许多样本的测量时间和成本。