Using an optimal CC-PLSR-RBFNN model and NIR spectroscopy for the starch content determination in corn

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Spectroscopy

Background:

• Corn starch is vital in food and chemical industries.
• Accurate and rapid starch content determination is crucial for quality control.
• Traditional methods may lack speed and reliability.

Purpose of the Study:

• To develop an enhanced methodology for rapid and reliable corn starch content determination.
• To introduce an optimal calibration model combining CC, PLSR, and RBFNN with NIR spectroscopy.
• To validate the model's predictive performance against established methods.

Main Methods:

• Utilized near-infrared (NIR) spectroscopy for spectral data acquisition.
• Developed an optimal calibration model by combining Correlation Coefficient (CC) method, Partial Least Squares Regression (PLSR), and Radial Basis Function Neural Network (RBFNN).
• Optimized crucial model parameters for enhanced accuracy.

Main Results:

• The proposed CC-PLSR-RBFNN model demonstrated superior predictive performance.
• Achieved the smallest Root Mean Square Error of Prediction (RMSEP) at 0.0497%.
• Obtained the highest Coefficient of Determination (Rp²) of 0.9968 on the prediction set.

Conclusions:

• The combined NIR spectroscopy and optimal CC-PLSR-RBFNN model offers a highly accurate and reliable method for determining corn starch content.
• This methodology significantly improves upon existing techniques for industrial applications.
• The study provides a valuable tool for efficient quality assessment in the corn industry.