Малогабаритный спектрометр ближнего инфракрасного диапазона для количественного определения состава древесины

A compact desktop near-infrared spectrometer with near-infrared spectrum acquisition software was developed based on fixed grating type fiber optic spectrometer with MEMS technology. The light source stability, baseline stability, instrument signal-to-noise ratio, and dark current drift are four important indicators for the performance test and evaluation of the spectral system. The test results show that the light source reaches a stable state after being warmed up for 2 s. The standard deviation of 100 %T line of1200-1550 nm instrument is less than 0.0003, and the signal-to-noise ratio is 3000:1. The dark current relative standard deviation fluctuates between 0.0019 and 0.0035. Based on 88 samples of the crushed material of lumber for wood cellulose and lignin contents, the quantitative calibration model was established using multiple scatter correction spectra pretreatment method set up after the correction model of cellulose and lignin. Root mean square errors of calibration set of cellulose and lignin are 0.6096 and 0.9572%, respectively. Root mean square errors of prediction set of cellulose and lignin are 1.2884 and 1.7712% respectively. The experimental results show that the developed small NIR spectrometer has a stable working state, and the prediction results for cellulose and lignin content based on the calibration model for powdery wood samples verify that the small NIR spectrometer has a high detection accuracy and can be used in the rapid detection of common materials.