Research on severe accident diagnosis method based on PCA and DT for a small modular PWR

Small modular pressurized water reactor (SMPWR) is a new trend in the development of nuclear energy today. SMPWRs usually adopt the integrated arrangement and use many passive safety systems, which have high inherent safety. However, despite all precautions, the severe accident (SA) cannot be completely avoided. It is necessary to conduct rapid and timely diagnosis for these SAs. Different from the traditional SA diagnosis methods based on signals and knowledge, the data-based method is adopted to study SA diagnosis of SMPWR in this study. First, the SA state monitoring method based on principal component analysis (PCA) is proposed to realize rapid and accurate distinction between steady-state and abnormal conditions. Then, the SA classification diagnosis method based on decision tree (DT) is proposed to realize accurate diagnosis of initial events and whether the entry conditions of SA management guideline (SAMG) are met. The key parameters selected through feature selection can provide supplement and reference for the selection of monitoring parameters in SAMG and the determination of the instrument list for instrument availability analysis under SA conditions. The fault diagnosis method proposed in this paper can provide the reference and basis for the SA diagnosis and the design of the operator's SA handling auxiliary system in the SMPWR.