[Application of PCA to diesel engine oil spectrometric analysis]

In order to study wear characteristics of a 6-cylinder diesel engine, six different working statuses were arranged by altering the clearance between cylinder and piston. Sixty-nine oil samples were taken from engine at different loads under 6 working statuses and analyzed by Spectroil M Instrument made in US. Principal component analysis (PCA) was applied to analyzing spectrometric data of sixty-nine oil samples and clustering those data according to elements and oil samples separately based on the weighted coefficient and principal component scores. All 21 elements were used in element clustering and only 6 wear-related elements, namely iron, chromium, aluminum, copper, plumbum and silicon, were used in sample clustering. It is shown that PCA effectively clustered oil spectrometric data into three different principal components according to elements. The projection of two different principal components exhibited five types of elements combinations, namely wear elements (Fe, Cr, Cu, Al and Pb), high concentration additives elements (Na, Zn, P, Ca and Mg), low concentration additives elements (Ba and B), base constituent of lubricating oils (C and H) and interferential elements (Ni, Ti, Mo, V, Ag and Sn). Furthermore, PCA clearly clustered oil samples according to different clearance between cylinder and piston in the diesel engine. The study suggests that analyzing oil spectrographic data by PCA could find the sources of different elements, monitor engine conditions and diagnose wear faults.