Monitoring and ANN Modeling of Coal Stockpile Behavior under Different Atmospheric Conditions

Abstract

In this study, an industrial-sized stockpile of 5 m width, 4 m height, and 10 m length was built in a coal stock area to investigate coal stockpile behavior under different atmospheric conditions. The effective parameters on the coal stockpile that were time, weather temperature, atmospheric pressure, air humidity, velocity, and direction of wind values were automatically measured by means of a computer-aided measurement system to obtain Artificial Neural Network (ANN) input data. The coal stockpiles, which should be continuously observed, are capable of spontaneous combustion and then causing serious economical losses due to the mentioned parameters. Afterwards, these measurement values were used for training and testing of the ANN model. Comparison of the experimental and ANN results, accuracy rates of training, and testing were found as 98.6% and 98.7%, respectively. It is shown that possible coal stockpile behavior with this ANN model is powerfully estimated.