Modeling approach to coal gasification using hybrid neural networks

Coal gasification was carried out in a bench-scale fluidized bed gasifier. The gasifier operated at atmospheric pressure with steam as fluidizing medium. The background of this study was the gas-steam cogeneration system. A hybrid neural network model was synthesized to predict the gas production rates. The model consists of a first principle partial model and a neural network parameter estimator. The model was trained with the experimental gasification data of Jincheng anthracite coal. The model has a good performance of process simulation. A parameter called 'fraction of active char' was proposed in the model. Neural network performed identification of this parameter and obtained its variation with temperature, with respect to Jincheng anthracite.

Overview