Rapid determination of moisture content of multi-source solid waste using ATR-FTIR and multiple machine learning...

Area of Science:

• Environmental Science
• Analytical Chemistry
• Data Science

Background:

• Accurate solid waste moisture content is crucial for effective recycling, treatment, and disposal.
• Traditional methods for moisture determination are slow, labor-intensive, and destructive.
• A rapid, non-destructive technique is needed for efficient waste characterization.

Purpose of the Study:

• To develop a rapid and nondestructive method for predicting the moisture content of multi-source solid waste.
• To evaluate the performance of combined spectral preprocessing and machine learning models for moisture prediction.
• To compare prediction accuracy using water-band versus full-band spectra.

Main Methods:

• Utilized attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) for spectral data acquisition.
• Applied multiple machine learning regression algorithms combined with various spectral preprocessing techniques.
• Optimized model hyperparameters and compared prediction results from different spectral regions.

Main Results:

• Achieved high prediction accuracy for moisture content in multi-source solid waste (textile, paper, leather, wood).
• The combined model demonstrated excellent performance with R² values of 0.9604 (validation) and 0.9660 (test), and a root mean square error of 3.80.
• The developed method proved efficient for rapid and accurate moisture content measurement.

Conclusions:

• The proposed ATR-FTIR and machine learning combined model offers a significant advancement for solid waste characterization.
• This technique enables real-time monitoring and improved management of solid waste treatment and disposal processes.
• The study highlights the potential of spectroscopic methods coupled with AI for sustainable waste management.