Research on the Assessment of Dairy Cow Dry Matter Intake Using ITSO-Optimized Stacking Ensemble Learning

Area of Science:

• Agricultural Science
• Machine Learning
• Animal Nutrition

Background:

• Dry matter intake (DMI) is crucial for dairy cow nutrition and production efficiency.
• Traditional DMI measurement methods are costly and complex.
• Existing neural network models for DMI assessment can be computationally intensive.

Purpose of the Study:

• To develop a cost-effective and accurate model for assessing dairy cow DMI.
• To optimize model parameters using an advanced metaheuristic algorithm.
• To enhance precision feeding strategies in dairy farming.

Main Methods:

• A Stacking ensemble learning model was proposed for DMI assessment.
• Model parameters were optimized using an improved Tuna Swarm Optimization (ITSO) algorithm.
• ITSO incorporated Sine-Logistic chaotic mapping, Levy flight, and Gaussian random walk for enhanced efficiency.

Main Results:

• The ITSO-optimized Stacking model achieved a high DMI assessment accuracy of 95.84%.
• The proposed model significantly outperformed other machine learning models (SVR, RF, DT, GBR, ETR, AdaBoost).
• Input variables included cow body weight, activity metrics, and feed ratios.

Conclusions:

• The ITSO-optimized Stacking model offers a robust and accurate solution for DMI assessment.
• This approach supports precision feeding, leading to optimized feeding strategies.
• The study contributes to improving dairy farm efficiency and promoting sustainable practices.