Development of thermodynamically assisted machine learning model to select best fuel for the thermal power station

Abstract

This work addresses a novel technique for selecting the best coal for a thermal power station using a thermodynamically assisted Machine Learning model. This work includes 32 coal samples in which the thermodynamically most suitable coal has been identified. Initially, the thermodynamic parameter ‘Quality’ has been derived from the primary constituents of coal. Five different Machine Learning algorithms have been used among which the Logistic Regression classifier was selected as the best because of the higher accuracy score. Feature selection has been carried out by using Random Forest and Logistic Regression methods. A python-based software for decision-making of coal feed has been developed where a multi-criteria decision making method (MCDM) includes Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) implemented to select the best coal.