Research on the online rapid sensing method of moisture content in famous green tea spreading

Moisture content is an important indicator in the process of green tea spreading. In order to realize the rapid detection of moisture content in the green tea spreading process, this paper collects 200 near‐infrared spectroscopy data during the spreading process and establishes a quantitative model of moisture content. To improve the performance of the model, standard normal variables (SNV), mean centering (MC), max–min normalization (Max–Min), detrending (Detrend), and Savitzky–Golay filtering (SG FIR) are used to improve the signal‐to‐noise ratio of the original spectrum. By comparing the partial least squares regression (PLSR) models established by the different pretreatment methods, SG FIR is determined as the best preprocessing method. To simplify the prediction model, the successive projections algorithm, competitive adaptive reweighted sampling, interval random frog, iterative retained information variables, variable combination population analysis (VCPA), and VCPA‐GA are used to extract the characteristic wavelengths from the preprocessed spectrum by SG FIR method. Then, based on the full spectrum and characteristic variables, the PLSR and support vector regression (SVR) models are established. The results show that established SVR prediction model based on VCPA‐GA extracting the characteristic wavelengths has the best effect. Among them, the correlation coefficients of the calibration set (Rc) is 0.9785, and the correlation coefficients of the prediction set (Rp) is 0.9658. Relative percent deviations are 3.6787. The results show that the VCPA‐GA + SVR model, based on near‐infrared spectroscopy technology, is the most effective and can accurately predict the moisture content in the green tea spreading process. PRACTICAL APPLICATIONS: At present, near‐infrared spectroscopy technology has been widely used in food detection. This study solved the problem of the inability to quantitatively predict the moisture content during the green tea spreading process by discussing different spectral preprocessing methods, characteristic wavelength selection methods and machine learning algorithms, and realized the rapid online detection of moisture during the green tea spreading process. This research has important guiding significance for the intelligentization of green tea processing equipment.