Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Sugiura, Yuki; Nonaka, Mai; Sugiyama, Yoshimi*; Hayashi, Keisuke*; Matsuura, Yasutaka*; Amano, Yuki; Ishidera, Takamitsu
no journal, ,
Crystalline rocks are being considered as potential host rocks for the geological disposal of high-level radioactive waste in several countries. The Kd values of radionuclides for crystalline rocks vary greatly and are complicated by multiple factors, such as solution conditions (e.g., pH, ionic strength), solid phase conditions (e.g., CEC, mineralogy), and the chemical properties of the radionuclides, making it difficult to predict Kd values under certain conditions. Recently, there has been an increase in the application of machine learning as a regression prediction method. In this study, we investigated the possibility of predicting Kd values of Cs for crystalline rocks using machine learning (XGboost) with the JAEA-SDB, which contains a large number of Kd values together with experimental conditions. As a result, we were able to predict the Kd values of Cs for crystalline rocks under a wide range of conditions, and to quantitatively evaluate the impact of each factor on Kd values by SHAP values.
Sugiura, Yuki; Nonaka, Mai; Sugiyama, Yoshimi*; Hayashi, Keisuke*; Matsuura, Yasutaka*; Amano, Yuki; Ishidera, Takamitsu
no journal, ,
no abstracts in English
Nagakubo, Azusa; Watanabe, Masatoshi*; Takubo, Kazuya*; Sugiyama, Yoshimi*; Hirouchi, Jun; Manabe, Kentaro; Takahara, Shogo
no journal, ,
no abstracts in English