Research Article
Partial Least Squares (PLS) Integrated Fourier Transform Infrared (FTIR) Approach for Prediction of Moisture in Transformer Oil and Lubricating Oil
Table 4
Moisture concentrations predicted with PLSR using the spectral data at 1700–1600 cm−1 and the measurements attained using Kittiwake method.
| | Expected concentration (ppm) | Predicted concentration in ppm (% error/CV) | | PLS regression (n = 5) | Kittiwake (n = 2) |
| | 0 | 68 ± 109 (NA/160%) | 500 ± 0 (160%/0%) | | 400 | 264 ± 357 (34%/135%) | 600 ± 0 (50%/0%) | | 600 | 850 ± 625 (42%/74%) | 550 ± 70 (8%/13%) | | 800 | 795 ± 114 (0.63%/14%) | 759 ± 70 (5%/9%) | | 900 | 892 ± 379 (0.9%/43%) | 700 ± 0 (22%/0%) | | 1200 | 906 ± 336 (25%/37%) | 850 ± 70 (29%/8%) | | 1800 | 1726 ± 234 (4%/14%) | 900 ± 0 (50%/0%) | | 2300 | 1962 ± 347 (15%/18%) | 1000 ± 0 (57%/0%) | | 2900 | 2936 ± 449 (1%/15%) | 1250 ± 70 (57%/6%) |
|
|
% error: percentage error; CV: coefficient of variation.
|
