Waste Plastic Color Prediction Using Pigment and Material Features

The patent changes the fundamental parameter of the prediction model from simple pigment addition to a machine learning-based model that incorporates waste plastic features. The prediction model learns complex non-linear relationships between pigment composition, waste plastic characteristics, and resulting color, enabling accurate predictions for recycled plastic materials where the simple addition rule fails.

The patent replaces the mechanical/simplified calculation approach (pigment addition rule) with an intelligent system based on machine learning. The prediction model automatically learns and captures complex relationships between multiple variables, substituting manual or formula-based methods with an adaptive computational system that improves accuracy without requiring complex manual interventions.

The prediction model performs self-learning from training data to automatically adapt to the complex interactions between pigments and waste plastic materials. By using the machine learning approach, the system self-adjusts its parameters and relationships without requiring complex external calibration or manual intervention, achieving reliable predictions for waste plastic materials while maintaining reasonable system complexity.