1. Study on irradiation embrittlement behavior of reactor pressure vessels by machine learning methods.
- Author
-
He, Wen-ke, Gong, Si-yi, Yang, Xin, Ma, Yan, Tong, Zhen-feng, and Chen, Tao
- Subjects
- *
PRESSURE vessels , *MACHINE learning , *NUCLEAR power plants , *EMBRITTLEMENT , *PRESSURIZED water reactors , *IRRADIATION , *COPPER - Abstract
• 9 chemical contents and 3 irradiation conditions were used as input factors. • The probability density distribution of Δ RT NDT was used to evaluated the model performance. • The GBDT model has the best performance, its prediction is close to that of ASTM E900-15. • FNF, T and the content of Cu, Ni, C are the most important 5 factors for irradiation embrittlement. A reliable irradiation embrittlement model of reactor pressure vessel (RPV) is critical to the safe operation and life extension of nuclear power plants (NPP). In this study, 474 datasets from NUREG/CR-6551 and 6 machine learning (ML) methods were adopted to accurately predict the irradiation embrittlement behavior (Δ RT NDT) of RPV with comprehensive input factors involving irradiation conditions and metal composition. The GBDT model has the best prediction performance. The calculated result by GBDT model is close to those of JEAC-4201 and ASTM E900-15. Cu content is the most sensitive factor for Δ RT NDT , followed by neutron fluence. This work demonstrates a successful application of ML in promoting NPP safety and RPV materials design. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF