Determining Sauter Mean Diameter Using Simple Optical Cell and Machine Learning Techniques

Abstract

Aerosol characterization plays crucial roles from environmental pollution measurements to fire safety technologies. The impact of aerosol mass concentration on human health is not fully captured by this environmental pollution measure alone. False alarms can occur with smoke detectors that solely rely on mass concentration data. This study estimated aerosol characteristics using Sauter mean diameter data through a combination of a basic optical cell and artificial neural network. The model, initially trained and tested using industrial standard materials, correctly predicted the values for new, untrained materials. The aim of this study was to transfer Sauter mean diameter information obtained from the literature without expensive calibration devices and complex optimization calculations to a measurement cell enhanced with machine learning technique. This study provides methods for obtaining Sauter particle diameter and mass concentration. It has versatile applications that range from particulate matter monitoring devices to smoke detectors.