The analysis of applicability of artificial neural networks as process models in chemical engineering taking as an example the process of drying in a fluidized bed dryer is presented. A number of neural models of the process, both classical and hybrid ones, were presented and tested. Dimensionless numbers were used as input and output data of the neural network, which allowed operating range of the network to be extended, while reducing the structure and increasing the dryer operation stability and accuracy at the same time.