Leakage Detection by Adaptive Process Modeling

Abstract-- In this paper, we propose an adaptive linear approach for time series modeling and steam line leakage detection. Weighted recursive least squares (WRLS) method is used for modeling. Interpretive variables of an adaptive model should be linearly correlated to ensure a robust model. In this paper it is ensured by examining eigenvalues and eigenvectors of the principal component analysis (PCA). The method is applied to a time series from the boiling water reactor (BWR) type nuclear power plant. Model is updated and used each time step to detect leakage in steam lines. Developed leakage detection index is based on the model estimation error. Method is more convincing in small pipe flows, because there are other ways to detect bigger volume leakages.

• Type: Conference
• Year: 2008
• Where: DMIN
• Authors: Jaakko Talonen, Miki Sirola, Jukka Parviainen
• Keywords: Data Mining, DMIN 2008, Leakage Detection, Steam Lines, Time Series