Study on the Determination of Flavor Value of Rice Based on Grid Iterative Search Swarm Optimization Support Vector...

Area of Science:

• Agricultural Science
• Food Science
• Spectroscopy

Background:

• Traditional rice flavor assessment relies on destructive chemical analysis and subjective evaluations, which are costly and time-consuming.
• Developing rapid, non-destructive methods for rice flavor analysis is crucial for efficient rice cultivation and processing.
• Existing methods lack the efficiency and user-friendliness required for modern agricultural practices.

Purpose of the Study:

• To develop and validate a novel non-destructive technique for determining rice flavor values.
• To improve the accuracy and efficiency of flavor assessment in various rice varieties.
• To introduce an optimized machine learning algorithm for hyperspectral data analysis in agriculture.

Main Methods:

• Utilized hyperspectral imaging technology for capturing spectral data from different rice varieties.
• Employed an improved Particle Swarm Optimization Support Vector Machine (PSO-SVM) algorithm, specifically the Grid Iterative Search Particle Swarm Optimization Support Vector Machine (GISPSO-SVM).
• Integrated feature extraction techniques, including Principal Component Analysis (PCA) and Competitive Adaptive Weighted Sampling (CARS), with the GISPSO-SVM model.

Main Results:

• The GISPSO-SVM algorithm demonstrated enhanced parameter finding ability compared to standard PSO-SVM.
• The combination of PCA and GISPSO-SVM achieved a high flavor value prediction accuracy of 96%.
• The GIS algorithm integration improved prediction accuracy across different feature selection methods, outperforming CARS (93% accuracy).

Conclusions:

• The developed hyperspectral imaging and GISPSO-SVM approach provides an effective, non-destructive method for rice flavor value detection.
• This novel technique offers a significant advancement over conventional methods, reducing resource depletion and costs.
• The study presents a new perspective for future research and application in rice quality assessment and breeding programs.