During the operation of nuclear power plants (NPP), the occurrence of severe accidents and critical equipment failures poses a significant threat to plant safety, creating stringent requirements for the accuracy and real-time performance of accident diagnosis and fault monitoring systems. Principal Component Analysis (PCA), as an effective dimensionality reduction algorithm, can extract key features from data, effectively reduce noise, and improve data processing efficiency. Consequently, PCA has been widely applied in recent years in nuclear power plant accident diagnosis and equipment fault monitoring. This paper systematically reviews the current applications of PCA techniques and their variants in NPP, focusing on the critical role of PCA and its variants in accident diagnosis and fault monitoring. Additionally, the paper explores the advantages of combining PCA with other machine learning techniques, such as Support Vector Machines (SVM) and neural networks, analyzing how these combinations enhance accuracy and robustness in accident diagnosis and fault monitoring. Finally, the paper summarizes the achievements and challenges of current PCA applications in nuclear plant diagnostic and monitoring systems and proposes potential future research directions.