Aiming at the measurement problem of crude oil moisture content

a detection system based on attenuated total reflection-Fourier transform infrared（ ATR-FTIR） spectroscopy is proposed. The principal component regression（ PCR） model with spectral data as input and the partial least squares model with spectral data and water content as input are established respectively to explore the model performance under the optimal preprocessing scheme. The results show that the analysis model of partial least squares（ PLS） combined with standard normal variate（ SNV）

SG convolution smoothing（ 7 points） and second derivative combined preprocessing method has the highest accuracy and prediction precision. The proposed method provides a good theoretical basis and technical support for the rapid detection of crude oil components and the supervision of crude oil quality in the industry.