Study on non-invasive classification of engine oil based on visible and short-wave near infrared spectroscopy

Visible and short-wave near infrared (Vis-SwNIR) spectroscopy was used for the non-invasive classification of engine oil. A total of 150 oil samples from three brands were prepared. The calibration set contains 120 samples which were randomly selected. The remaining 30 samples were used for the prediction. After the spectra measurement, principal component analysis was calculated to cluster the samples. Discrete wavelet transform (DWT) was used to do the spectral mining. The obtained wavelet coefficients were inputted into artificial neural network (ANN) for the brand classification of engine oil. The correct classification rate of 100% was obtained by DWT-ANN model. The overall results show that Vis-SwNIR spectroscopy is a feasible technique for the brand classification of engine oil.