Rapid Determination Of Oxides Content In Cement Raw Meal Based On Near Infrared Spectroscopy

The contents of four oxides in cement raw meal were determined by near infrared spectroscopy. In this method, the near infrared spectrum was collected by diffuse reflectance method, and X-ray fluorescence spectroscopy analysis was used to determine the reference values of oxides content. The outliers were eliminated according to the Mahalanobis distance, and then the samples were divided into calibration subset and validation subset by sample set partitioning based on joint X-Y distance (SPXY) method. The optimal wavelength variables were selected by backward interval partial least squares and genetic algorithm. Quantitative calibration models of four oxides were established by partial least squares algorithm, which showed good prediction performance. The correlation coefficients of validation subsets of CaO, SiO2, Al2O3 and Fe2O3 models were 0.9411, 0.9337, 0.8612 and 0.7351, the root mean square error of prediction were 0.0994, 0.1044, 0.0693 and 0.0387, the average absolute errors were 0.075%, 0.083%, 0.051% and 0.025%, respectively. By comparing with prompt gamma-ray neutron activation analysis and laser-induced breakdown spectroscopy analysis, the near infrared spectroscopy analysis has short time-consuming, highest accuracy, and best effect. It is suitable for rapid and accurate determination of contents of four oxides in cement raw meal, and provides a new idea for quality monitoring of cement raw meal.