Machine Learning-Based Process Control for Injection Molding of Recycled Polypropylene

Area of Science:

• Manufacturing
• Materials Science
• Artificial Intelligence

Background:

• Recycled materials in manufacturing, especially injection molding, present challenges due to variability affecting part quality and process stability.
• Artificial intelligence (AI) and machine learning (ML) offer potential solutions for optimizing manufacturing processes and improving efficiency.
• Addressing material variability is crucial for the successful integration of recycled plastics into industrial applications.

Purpose of the Study:

• To develop and evaluate a novel closed-loop process control approach for injection molding using machine learning.
• To adaptively predict processing inputs and quality outcomes for recycled polypropylene (rPP) blends.
• To assess the effectiveness of different machine learning models in handling material variability.

Main Methods:

• Utilized five blends of recycled polypropylene (rPP) for testing.
• Employed artificial neural networks (ANNs), linear regression, and polynomial regression to model relationships between material properties and process parameters.
• Implemented an ANN with TensorFlow and Keras, featuring six hidden layers, ReLU activation, and Adam optimizer, with empirical tuning and early stopping.

Main Results:

• Yield stress, ultimate elongation, and part weight were predicted with high accuracy (within 5% error for regression models, 10% for ANN).
• Modulus predictions showed higher variability (errors ~11% for ANN/linear, ~40% for polynomial), reflecting rPP's inherent property fluctuations.
• Processing input predictions had errors ranging from 3% to 25%, varying by model and response.

Conclusions:

• Closed-loop process control powered by machine learning effectively predicts quality parameters in injection molding of recycled materials.
• The study highlights that no single modeling approach consistently outperforms others, underscoring the complexity of material-process-quality relationships.
• The proposed approach can enhance process stability, improve material utilization, and promote the adoption of sustainable materials in manufacturing.