Ozsen, SeralOzsen, HakanSensogut, Cem2020-03-262020-03-2620111314-2704https://hdl.handle.net/20.500.12395/2630511th International Multidisciplinary Scientific GeoConference -- JUN 20-25, 2011 -- Albena, BULGARIACoal is the most consumed fossil fuel in the world. Determination of the thermal properties of coal is a very important matter and it is not straightforward because of the heterogeneous structure of the coal. The short and elementary analysis results of coals with different carbonization degrees are different. The mineral composition of a coal also affects the thermal behavior. To detect thermal properties of coals, thermal analysis devices are generally used in many widespread methods. The most widely used methods in thermal analysis of coals are Differential Thermal Analysis (DTA) and Thermogravimetry (TG). In this study however, a different analysis method to determine combustion degree of coals was applied. By utilizing from some properties of coals obtained by short analysis and sulphur analysis, an Artificial Neural Network (ANN) was trained to predict the combustion degrees of coals. For this application 84 coal samples were prepared from 28 different locations in TURKEY. Among these, 67 samples were used in training ANN and the remaining 17 were used in test procedure. For the test samples, the trained ANN was used to predict the combustion degrees of them by presenting 8 different properties obtained from short and Sulphur analysis results. Then the mean squared error (mse) was calculated between the real combustion degrees which were also determined from the TG method and predicted combustion degrees of ANN. The test mse was found to be 2.9x10(-4). This result means that the trained ANN could predict combustion degree of a coal sample with a mean error of 2.9x10(-4). When the time and effort spend on determining thermal property of a coal sample with a classical method is considered, this gives another alternative to the experimenter for determining combustion degree of that sample in more short and effortless manner.eninfo:eu-repo/semantics/closedAccessTGDTAcoalcombustion propertiesartificial neural networks (ANN)Conference Object759+N/AWOS:000307366000101N/A