Efficient algorithmic coupling technique for precision recycling of seven types of mixed plastic waste

The annual global production of plastic waste, characterized by complex composition and challenges in separation, necessitates immediate and comprehensive measures for the recycling and disposal of mixed plastic waste in an environmentally friendly and meticulous manner. This study introduces an efficient two-step coupling technique, employing Linear Support Vector Classification (Linear-SVC) in tandem with Multi-layer Perceptron (MLP). The application of this coupling technique elevates the overall accuracy of identifying seven types of plastics from 94.7% to an impressive 97.7%. Furthermore, the method exhibits a reduced running time compared to the one-step method of MLP. Notably, the classification accuracy for high-density polyethylene (HDPE) and low-density polyethylene (LDPE) experiences a substantial improvement from 79–94%, outperforming the one-step MLP method. This coupling technique emerges as an effective strategy, contributing significantly to the harmless and precise recycling of waste plastics.