Application Of Quantitative Data-Based Fault Detection Methods on a Drum-Type Boiler

In this paper we present methods of selecting the optimum number of principal component analysis (PCA). The optimal number of PCs can be selected via the most valuablesingular value or PRESS method and the performance of fault detection can be improved. These methods are applied to data of a power plant drum-type boiler. The m-file results demonstrate its good performance

کلیدواژه ها:

Fault detection ، Drum boiler ، State space ، Principal component analysis