Some additional results on principal components analysis of three-mode data by means of alternating least squares algorithms

Psychometrika ◽  
1987 ◽  
Vol 52 (2) ◽  
pp. 183-191 ◽  
Author(s):  
Jos M. F. ten Berge ◽  
Jan de Leeuw ◽  
Pieter M. Kroonenberg
Keyword(s):  
Least Squares ◽  
Principal Components Analysis ◽  
Principal Components ◽  
Alternating Least Squares ◽  
Components Analysis

Temperature-Induced Chemical Changes in Lubricant Automotive Oils Evaluated Using Raman Spectroscopy

2020 ◽  
pp. 000370282094994
Author(s):  
Andressa Cristina de M. Bezerra ◽  
Nelize Maria de A. Coelho ◽  
Felipe Bertelli ◽  
Marcos Tadeu T. Pacheco ◽  
Landulfo Silveira
Keyword(s):  
Raman Spectroscopy ◽  
Least Squares ◽  
Principal Components Analysis ◽  
Partial Least Squares ◽  
Principal Components ◽  
Heating Time ◽  
Chemical Changes ◽  
Lubricating Oils ◽  
Spectral Models ◽  
Components Analysis

Automotive engine lubricating oils are not only intended to reduce friction between parts, but also act on the cooling of motor components and protection of metals against corrosion. To improve its properties and efficiency, additives are added to the base oil for different goals. However, over time of use, external factors modify its properties, such as the engine operating temperature, the frictional force between parts, the mixture of this oil with fuel before burning and with combustion products, causing loss of their efficiency. This work aimed to evaluate, with Raman spectroscopy technique, the temperature-induced changes related to degradation of mineral, semi-synthetic and synthetic automotive lubricating oils. Samples being subject to periodic heating cycle were kept to average temperature of 133 ℃, considering 8 h per day, for six days, until complete 48 h of heating. By analyzing the Raman spectra, it was possible to identify common peaks between the three types of oils and changes caused by heating cycles. Principal components analysis showed that the synthetic oil degraded in less extent than the semi-synthetic one, and this one degraded less than the mineral oil. Spectral models to predict the heating time based on the spectral variations identified using principal components analysis and the regression done using partial least squares, using the heating time as independent variable and the spectral features as dependent variables, was able to predict the heating time for each of oil types with high correlation and prediction error ( r > 0.97 and error <4.0 h) for both principal components analysis and partial least squares regression models. Raman technique was able to identify chemical changes resulting from the heating of lubricant oils and to correlate these changes with the heating time, thus becoming a technique of interest for the preventive maintenance area.

Sign in / Sign up

Export Citation Format

Share Document