Forecasting of Ra(226), Th(232) and U(238) Concentrations using Artificial Neural Networks (ANNs)

Abstract

Identification and modeling of radioactive concentrations in a region is very important for the region in terms of radiological hazards. Artificial Neural Network (ANN) can successfully model large systems. The validity of the model was tested by entering the data of the proposed ANN model that had never been entered into the system. In this research, average activity concentrations of 226Ra, 232Th and 238U in the water samples collected from the lake are 1.439 Bql-1, 4.508 Bql-1 and 14.682   Bql-1, respectively. The characteristics of the study area are also determined with the spatial maps and ANNs are used to prediction and modeling of the radionuclides. The mean square errors for the obtained results are less than 1.5%. The correlation coefficient close to +1 indicates the validity of the model for this study.