The Potentialities of Machine Learning for Cow-Specific Milking: Automatically Setting Variables in Milking Machines.

Area of Science:

• Agricultural Engineering
• Animal Science
• Artificial Intelligence

Background:

• Milking operations are critical for dairy farm profitability but are time-consuming and costly.
• Reducing milking duration without compromising animal welfare is a key objective for dairy management.

Purpose of the Study:

• To evaluate the impact of varying pulsation ratios and detachment flow rates on milking performance and cow health.
• To develop an optimized artificial intelligence model for predicting cow-specific milking parameters.

Main Methods:

• Two trials were conducted on a dairy farm, testing different pulsation ratios (60:40 vs. 65:35) and detachment flow rates (600 g/min vs. 800 g/min).
• An optimized least-squares support vector machine (LSSVM) classification model, enhanced by the sparrow search algorithm (SSA), was developed.
• The model predicted optimal pulsation ratio and detachment flow rate for individual cows within two minutes of milking initiation.

Main Results:

• The LSSVM-SSA model demonstrated high accuracy (92%) and precision (97%) in shortening milking times based on pulsation ratios.
• The model achieved 78% accuracy and 79% precision for predicting optimal detachment flow rates.
• Evaluated parameters included milking performance, somatic cell counts, clinical mastitis incidence, and teat scores.

Conclusions:

• Implementing the developed AI algorithm in non-automatic milking machines can enable cow-specific milking operations.
• Personalized milking parameters have the potential to significantly reduce milking time and improve overall farm efficiency.