Research on Fault Detection and Identification Method of Small PWR Based on Principal Component Analysis

Fault detection and identification are important for the safety and economy of small PWRs. The fault detection and identification method based on signal and expert knowledge and experience is usually applied in nuclear reactors. However, operators are often unable to identify the fault type and trace the fault cause in time and accurately from the massive fault data information. A method of fault detection and identification of small PWR based on principal component analysis is presented in this paper. First, the model of a small PWR is established by RELAP5 code, and the sample data of typical faults is obtained. Second, the dimension of sample data is reduced by using principle component analysis method. T² and Q statistics are calculated to detect the reactor operation condition by judging whether the thresholds are exceeded. Then, the contribution rate of process variables to statistics is analyzed by using the contribution rate graph method, so as to determine the variables that play a major role in the change of fault characteristics and realize the identification of different faults. Finally, the effectiveness of the method is verified by comparing with the actual physical process analysis results.