The principal components analysis and Hotellings T2 statistic are used in studying the clinker minerals of Portland Cement Company of Northern Nigeria (CCNN). The cement produced is considered to be a mixture of eight minerals, each being sensitive to the presence of fly ash. Effort was made in this paper to show that the dependence on only one mineral in detecting fault or diagnosing noise was not effective but rather using the entire components such that each components contribution to variation is measured thereby making a valid interpretation about fault.
Key words: Hotellings T2 approximated model, Fault detection, PCA, Components treatments
|
