Author ORCID Identifier

https://orcid.org/0009-0009-3344-4621

Date Available

8-20-2025

Year of Publication

2025

Document Type

Master's Thesis

Degree Name

Master of Science in Biosystems and Agricultural Engineering (MSBiosyAgE)

College

Agriculture; Engineering

Department/School/Program

Biosystems and Agricultural Engineering

Faculty

Dr. Jian Shi

Faculty

Dr. Michael Sama

Abstract

According to the U.S. Department of Energy’s 2023-billion-ton report, the United States generates nearly 300 million tons of municipal solid waste every year. More than 50% of that waste is underutilized and ends up in landfills. This poses serious public health and environmental risks. But beyond the environmental and health implications, this is also a wasted opportunity. The U.S. Energy Information Administration estimates that, for every 100 lbs of waste we throw away, about 85% can be used as fuel or for electricity generation. However, its complex and heterogeneous nature poses challenges for efficient sorting and valorization in waste to energy systems. In this project, we employed hyperspectral Near-Infrared (NIR) imaging, with various preprocessing techniques for the detection and classification of various MSW fractions (paper, cardboard and various waste plastics) collected from homes in Lexington, Kentucky. Simultaneously, comprehensive chemical and structural analyses, including ultimate and elemental analysis, Fourier Transform Infrared (FTIR) spectroscopy, and Scanning Electron Microscopy (SEM), were conducted. This study successfully developed and evaluated machine learning models, including Support Vector Machines (SVM), Random Forest (RF), k-Nearest Neighbors (KNN), and a 1D Convolutional Neural Network (1D-CNN), for automated classification based on the hyperspectral data. The results showed accuracies above 90% across all models. The classification was also performed on selected wavelengths of 3, 5, 10, and 20 for both Recursive feature elimination and sequential forward selection. Overall, the results indicate that NIR hyperspectral imaging could be used to classify different waste materials rapidly.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2025.434

Funding Information

This work was funded by the United States Department of Energy’s Bioenergy Technologies Office (Award Number: DE-EE0010295)

Share

COinS