Determination of diesel properties by infrared spectroscopy and multivariate calibration

Identificador:  TDX:4490

Handle: https://hdl.handle.net/20.500.11797/TDX4490

Autores:  Rodríguez Barrios, María Suliany

Resumen:
In this doctoral thesis, multivariate calibration models based on infrared (IR) spectroscopy are developed to predict eleven quality properties of diesel. To develop the models, a set of commercial and desulfurized diesel samples collected over almost four years have been analyzed. Calibration models based on artificial neural networks (ANN) and partial least squares (PLS) regression have been considered to model the relationships between the IR spectrum and properties. For the ANN model, a new strategy has been developed to define the limits of its domain of applicability. PLS and ANN models accurately predict density, cetane number, FAME content and viscosity. Three adaptations of the PLS calibration model are explored to predict the density of commercial diesel samples when it is used with samples analyzed in an instrument different from the one used to develop the model. The adaptations used were diPLS, DOP and MU.