Study on Spectral Preprocessing and Wavelength Selection for Determining Kinematic Viscosity of Aviation Lubricating Oil By Infrared Spectroscopy

The effects of spectral preprocessing and wavelength selection on determining kinematic viscosity of aviation lubricating oil by infrared spectroscopy were studied. The methods were optimized respectively and compared with each other by the correction coefficients and root mean square error of the calibration set and validtion set. The noise filtering effect of wavelet transformation method is better than Savitzky-Golay smoothing convolution in spectral preprocessing of aviation lubricating oil. The prediction accuracy of the model with daubechies2 wavelet function and 7 decomposition levels was highest. Meantime, the application effects of three spectral wavelength selection methods, including correlation coefficient, backward interval partial least squares and synergy interval partial least squares were studied. The spectral wavelength selected by the synergy interval partial least squares with 4 combination intervals in 37 subspace intervals was the best to model prediction effect. The wavelet transformation combined with difference derivation and was applied in spectral preprocessing, and the model on kinematic viscosity of aviation lubricating oil was built by siPLS. The ten used oil were analyzed by the model. The experiment results demonstrate that model predict kinematic viscosity of samples accurately and stably.