A Novel Genetic Algorithm-Based Optimization Framework for the Improvement of Near-Infrared Quantitative Calibration...

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Biotechnology

Background:

• Fishmeal is a key protein source in animal feed, necessitating accurate nutritional monitoring.
• Near-infrared (NIR) spectroscopy offers rapid analysis but requires robust calibration models.
• Improving NIR quantitative calibration is crucial for precise protein content determination in feed ingredients.

Purpose of the Study:

• To develop and validate a novel optimization framework (GSMW-LPC-GA) for enhanced NIR quantitative calibration of fishmeal protein.
• To investigate the effectiveness of combining spectral selection and chemometric optimization for improved prediction accuracy.
• To establish an efficient strategy for analyzing nutritional components in animal feed.

Main Methods:

• A novel GSMW-LPC-GA framework was developed for NIR calibration.
• Informative NIR wavebands were selected using the Grid Search Moving Window (GSMW) strategy.
• Latent Principal Components (LPCs) were derived from selected wavebands for input into a Genetic Algorithm (GA) for secondary optimization.

Main Results:

• The GSMW-LPC-GA framework demonstrated superior NIR prediction performance for fishmeal protein compared to conventional moving window models.
• Genetic Algorithm (GA) optimization effectively improved the performance of NIR calibration models.
• The proposed framework proved suitable for quantitative determination of fishmeal protein content.

Conclusions:

• The GSMW-LPC-GA framework provides an efficient and effective strategy for improving NIR quantitative calibration models.
• This approach enhances the accuracy of protein analysis in fishmeal, crucial for animal husbandry.
• The methodology is adaptable for analyzing nutritional changes and influences in various feed components.