A comparative study of diesel analysis by FTIR, FTNIR and FT-Raman spectroscopy using PLS and artificial neural network analysis

摘要: Abstract Diesel properties determined by ASTM reference methods as cetane index, density, viscosity, distillation temperatures at 50% (T50) and 85% (T85) recovery, the total sulfur content (%, w/w) were modeled FTIR-ATR, FTNIR, FT-Raman spectroscopy using partial last square regression (PLS) artificial neural network (ANN) spectral analysis. In PLS models, 45 diesel samples used in training group other validation. ANN analysis a modular feedforward was used. Sixty 30 Two different ATR configurations compared FTIR, conventional (ATR1) an immersion (ATR2) cell. The ATR1 cell presented best results, with smaller prediction errors (root mean error of prediction, RMSEP). comparison three models (FTIR-ATR1, FT-Raman) shows that reasonable values R2 RMSEP obtained FTIR-ATR1 FTNIR evaluation T50. PLS/FT-Raman results only for T50 property. None techniques able to generate suitable calibration determination content. ANN/FT-Raman performances, all presenting R2-values above some them significantly than those FTIR-ATR FTNIR. ANN/FTIR-ATR1 estimate 0.01% (w/w) accuracy.