Prediction of product distribution of low-medium rank coal pyrolysis using artificial neural networks model

Abstract

In this study, artificial neural networks (ANNs) were used to build a correlation between the low-medium rank coal elements and the coal pyrolysis process parameters and product distribution. On the basis of numerous experimental data of different rank coals under various process conditions, a network training database was created in order to more correctly analyze the product distribution during coal pyrolysis. One ANN model was created for each of the six pyrolysis products, for a total of six models. The validation database validates the accuracy and applicability of the constructed ANNs model, and it also assesses the relative influence of each input parameter on the evolution of each pyrolysis product. The results show that the predicted values for the new inputs in the developed network model have strong correlation coefficients and are in good agreement with the experimental values of pyrolysis products, demonstrating its accuracy and robustness. Thus, it is demonstrated that the ANN based methodology is a viable alternative which can be used to predict the product yield of coal pyrolysis.