Detection of Soluble Solids Content in Different Cultivated Fresh Jujubes Based on Variable Optimization and Model...

Area of Science:

• Agricultural Science
• Spectroscopy
• Data Analysis

Background:

• Visible/near infrared (VIS/NIR) spectroscopy models often fail for soluble solids content (SSC) detection in fresh jujubes.
• Cultivation methods (open field vs. rain shelter) can impact fruit quality and model accuracy.

Purpose of the Study:

• To develop a robust VIS/NIR spectroscopy model for accurate SSC detection in fresh jujubes.
• To address the failure problem of existing models by optimizing variable selection and updating the model.
• To enable reliable identification of jujubes based on cultivation environment.

Main Methods:

• Variable optimization using Iteratively Retained Informative Variables (IRIV) and Successive Projections Algorithm (SPA).
• Establishment of detection models using Least Square Support Vector Machine (LS-SVM).
• Wavelength fusion and recalibration using Euclidean distance for model updating.

Main Results:

• The IRIV-SPA method outperformed IRIV for characteristic wavelength extraction.
• The updated LS-SVM model achieved good prediction results for SSC.
• The model demonstrated high accuracy for open-field (R²=0.82, RMSE=1.49%, RPD=2.18) and rain-shelter (R²=0.81, RMSE=1.44%, RPD=2.17) grown jujubes.

Conclusions:

• A novel wavelength fusion-Euclidean distance update method enhances VIS/NIR model robustness for fruit quality detection.
• This approach effectively addresses the need for accurate SSC detection in fresh jujubes across different cultivation environments.
• The study provides a valuable method for establishing reliable fruit quality assessment models.