Machine Learning Regression Model in MATLAB for Predicting Gas Emissions During the Pellet Combustion

Machine learning is a very popular tool in data analysis techniques. Its algorithms use computational methods to learn information directly from data. The following two techniques can be used: supervised learning (classification and regression) and unsupervised learning (clustering). This chapter deals with regression in MATLAB, which trains a model on known input and output data so that it can predict future outputs. It created a regression model for the prediction of the formation of gas emissions (carbon monoxide and nitrogen oxides) during pellet combustion. Selected data were trained on spruce samples with 2%, 3%, 20%, and 50% of the content of face masks and validated on a spruce pellet sample with 10% of the content of face masks. Disintegrated face masks were added to spruce sawdust and compressed into pellets. Another benefit of this chapter is the disposal, further processing, and use of this waste. Gas emissions were measured by a flue gas analyzer in the chimney of a small heat source. Other quantities such as temperature, pressure, and airflow were recorded and used as input data in the model. This tool allows emissions to be predicted without their measurements during the combustion of solid fuels such as pellets.