| Abstract:In this study， the feasibility of near-infrared （NIR） modeling using an industrial near infrared spectrometer for rapid prediction of wood chips properties was explored and compared with the modeling results of experimental spectrometer. According to the characteristics of different wood properties， the model performance was optimized by spectral signal correction and wavelength screening algorithms， and a suitable modeling method was established. The results showed that the NIR model based on an industrial spectrometer exhibited higher prediction accuracy for physical properties such as moisture， basic density， and fiber length of wood chips， while the prediction accuracy of chemical component content was not significantly different from that of the experimental spectrometer. NIR model of wood chip properties based on industrial spectrometer was suitable for raw material quality evaluation and industrial process analysis. |
