Establishment of Near-Infrared Prediction Models for Toluene Insoluble Substance and Other 3 Indexes of Medium Temperature Coal Pitch

Near-infrared diffuse reflectance spectra of medium temperature coal pitch samples were collected with integrating sphere analysis mode.SG1D, SNV and their combined methods were applied for pretreatments of the original spectra. Determined results of 4 indexes including TI content, QI content, softening point and coking value were obtained by GB/T 2292-2018, GB/T 2293-2019, GB/T 2294-2019 and GB/T 8727-2008. With values of NIR spectrum as independent variable and determined values as dependent variable, prediction models of 4 indexes were established by PLS. As found by results that the correlation between the predicted values and determined values was good, giving values of R² of 0.82, 0.79, 0.71, 0.96, and values of RMSEC found were 0.48%, 0.35%, 1.11℃, 0.35%, with values of RMSEP of 0.59%, 0.45%, 1.10℃, 0.52%, respectively. The model had a good predictive ability and could be applied in the quality control of the manufacturing process of coal pitch.