Coal is still a significant energy source for the world. Due to the utilization of low-rank coal, drying is a key issue. There are lots of attempts to develop efficient drying processes. The most prominent method seems as thermal drying. For thermal drying processes, the most important subject is the coal moisture content change with time. In this study, convective drying experiments were utilized to develop a new model based on decision tree regression method to predict coal moisture content. The developed model gives satisfactory results in prediction of instant coal moisture content with changing drying conditions. With the decision tree depth of six, the best test results were achieved as 0.056 and 0.802 for MSE and R² analyses, respectively.

Coal is an important component in the energy industry and plays a key role in energy-producing facilities. Moisture is a common condition that has a considerable impact on coal. Coal drying has long been a question of great interest in a wide range of fields. Defining parameters in the coal drying is obtained by experiments. High costs, time constraints, and repetition of an experiment are one of the most frequently stated problems with experimental works. Using qualitative methods with experiments can be more useful for identifying and characterizing the coal drying process. The purpose of this article is finding the effective parameters in the coal drying process by using a hybridized prediction method. Genetic Algorithm (GA) and Artificial Neural Network (ANN) are hybridized with each other to identify and characterize the coal drying process. GA-ANN algorithm is applied to the coal drying process to predict the moisture of coal, but it does not provide a decent result at first. Later, the Design of Experiment (DoE) methodology is performed to determine the main effects of six parameters. Two scenarios are generated because two parameters are not statistically significant. The first scenario excludes the air relative humidity parameter, and the second scenario excludes the air relative humidity and the velocity of air parameters. Following the application of the DoE method, GA-ANN reaches decent results in scenario-2.