Diagnosis of defects by principal component analysis of a gas turbine

Abstract

This study examines the application of the Principal Component Analysis (PCA) technique to detect the failures in complex industrial processes such as gas turbines used for electric power generation. The early detection of failures in such complex processes is indeed paramount to prevent product deterioration, performance degradation, significant property damage and human health. We identified the PCA model by determining the optimal number of principal components retained in the PCA model, then we validated the PCA model by checking the evolution of measurements and estimated the two variables X2 and X8. Thereafter, the evolution of three detection indexes is illustrated highlighting that the filtered SPE index is the best suited one for our installation, and finally, we checked the efficiency of the linear PCA method from the filtered SPE detection index using real data of defects that may occur within the gas turbine. The results obtained will aid to confirm the performance of the linear PCA method in the field of early failure detection. Thus, the PCA method appears as an efficacious tool to monitor and diagnose complex installations.