Effect of fruit moving speed on online prediction of soluble solids content of apple using Vis/NIR diffuse transmission

摘要

The effect of fruit moving speed on online prediction of soluble solids content (SSC) of Fuji apples based on visible and near-infrared (Vis/NIR) spectroscopy was studied. Diffuse transmission spectra between 615 and 1,045nm were collected with a commercial online system at speeds of 0.3 m/s (S1), 0.5 m/s (S2), and 0.7 m/s (S3). Compensation models for SSC of each speed alone (local models) and all speeds (global model) were established using partial least squares (PLS). For global model, spectra of each sample were divided into three parts (P1, P2, and P3), three kinds of spectra partition combinations (P12, P13, and P23) were established. Results showed that S3 performed better and the influence of speed on spectra greatly affected SSC evaluation accuracy between local models. Comparatively, global model was insensitive to fruit moving speed variation and effective wavelengths (EWs) selected by competitive adaptive reweighted sampling (CARS) after Savitzky-Golay smoothing (SGS) achieved better results than local models. Importantly, 36 EWs selected by CARS after SGS of global-P13 model achieved the best results with r(p) and RMSEP of 0.8419, 0.8895, 0.8948 and 0.6281, 0.5318, 0.5196 degrees Brix, respectively. Generally, global-P13 model with EWs is promisingly applied to online SSC prediction of apple by Vis/NIR diffuse transmission.Practical applicationsIt is of great value to study nondestructive and rapid methods for detecting apple SSC to meet the ever-growing needs of consumers for continuous supply and high-quality fruit. The methods and results in this study can provide basic references for Vis/NIR diffuse transmittance online detection of SSC of Fuji apple about fruit moving speed variation and model simplification. It is of great social value for developing a real-time evaluation system of apple SSC in postharvest treatment or focusing on quality changing in storage shelf.