Scholarlink

<p>Plastic waste management remains challenging because different polymer types are often difficult to distinguish during sorting. This study presents a plastic classification system that combines a Near-Infrared (NIR) sensor with machine learning for automated identification. Spectral reflectance data were collected from polypropylene (PP), polyethylene terephthalate (PET), low-density polyethylene (LDPE), and baseline samples. Multiple machine learning models were evaluated, and the best-performing methods were combined to improve classification performance. The system was implemented using an Arduino-based sensing setup and a Python-based machine learning pipeline. Experimental results demonstrated an accuracy of over 98% during real-time testing, indicating the potential of the proposed low-cost system for automated plastic sorting and waste management applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A hybrid K-Means–KNN approach for real-time plastic type identification using NIR spectroscopy

  • 期刊论文

  • 发表日期：

    2026年8月1日

查看全文
  • Abijjith VR,
  • Ananya,
  • Prakash Marimuthu

摘要

Plastic waste management remains challenging because different polymer types are often difficult to distinguish during sorting. This study presents a plastic classification system that combines a Near-Infrared (NIR) sensor with machine learning for automated identification. Spectral reflectance data were collected from polypropylene (PP), polyethylene terephthalate (PET), low-density polyethylene (LDPE), and baseline samples. Multiple machine learning models were evaluated, and the best-performing methods were combined to improve classification performance. The system was implemented using an Arduino-based sensing setup and a Python-based machine learning pipeline. Experimental results demonstrated an accuracy of over 98% during real-time testing, indicating the potential of the proposed low-cost system for automated plastic sorting and waste management applications.