Application of PCA for early leak detection in a pipeline system of a steam boiler

摘要

The application of the Principal Component Analysis (PCA) method for early detection of leakages in the pipeline system of a steam boiler in a thermal-electrical power plant is presented and discussed. The PCA model built from historical measurements of 12 selected process variables, mapped to the reduced space of three Principal Components (PC) of the highest magnitude, was used to establish the confidence ellipsoid, i.e. the feasible region in the PC coordinates, occupied by the values of process variables related to the 'healthy' system. Changes of the current location of the process operating point in the PC space created the 'fault trajectory' and were the basis for making a decision of leakage detection.