Abstract: The internal circulation of particles in a dual circulating fluidized bed directly determines the heat transfer effect in the sub fluidized bed. In order to understand the heat transfer effect of double fluidized bed under other conditions,a variety of ANN models were used to predict the internal circulation of particles. Based on the data from results of 60 sets of operating conditions with a new dual fluidized bed cold state experimental system,five different artificial neural network（ ANN） models were used to predict the internal circulation of particles,and the experimental data were compared with each other. The flow velocity,the height of furnace partition wall and the initial bed height of the main stream bed were used as the characteristic values of the model. The internal circulation of particles was taken as the output target,and the average absolute percentage error and root mean square error were used as evaluation indexes. The result shows that GA-BP,SVM and elm have ideal prediction accuracy.