Data reconciliation and gross error analysis of self powered neutron detectors: comparison of PCA and IPCA based models

Abstract

Self powered neutron detectors (SPNDs) are used for measuring neutron flux in a nuclear reactor and hence are essential for safe operation of the reactor. The main objective in this paper is to develop models on representative groups of SPNDs using principal component analysis (PCA) based techniques. In particular, we identify models using regular PCA and the iterative PCA (IPCA) technique proposed in Narasimhan and Shah (Control Eng Pract 16:146–155, 2008). These models are then used for performing data reconciliation, gross error detection, identification and estimation for the SPNDs. Based on these performances, we compare the PCA and IPCA models. The proposed models can be used in the reactor to ensure that faulty detectors can be detected quickly and the corresponding values estimated in real-time thereby ensuring continuous reactor operation.