Parameter optimization of double-blade normal milk processing and mixing performance based on RSM and BP-GA

Area of Science:

• Food Engineering
• Process Optimization
• Computational Modeling

Background:

• Optimizing milk processing is crucial for product quality and efficiency.
• Traditional methods may not fully capture complex interactions influencing processing performance.

Purpose of the Study:

• To optimize the processing performance of a double-blade normal milk mixer.
• To compare the effectiveness of Response Surface Methodology (RSM) and a BP-GA neural network algorithm for process optimization.

Main Methods:

• Utilized temperature stability as the key performance indicator.
• Employed Response Surface Methodology (RSM) for initial optimization.
• Applied a BP-GA (Backpropagation-Genetic Algorithm) neural network for advanced optimization.
• Identified key factors: blade shape, height, and rotating speed.

Main Results:

• Blade shape was the most influential factor on temperature stability, followed by height and rotating speed.
• The BP-GA neural network achieved a lower relative error (2.9%) compared to RSM (4.5%).
• The BP-GA model demonstrated superior fitting performance with a determination coefficient of 0.9960.

Conclusions:

• The BP-GA neural network algorithm offers better fitting performance for optimizing milk processing parameters.
• The study identified optimal working parameters for the double-blade normal milk mixer.
• Findings provide valuable insights for enhancing milk processing device design and overall milk quality.