The problem of the use of neural networks in the control system for the production of viscoelastic media is considered. Classical methods of process control are considered. A new approach to the process control of using neural networks is offered. As an example, the problem of control product concentration at the outlet of the catalytic reactor into a two-layer feedforward neural network is solved. Quality criterion of optimal control algorithm with the prediction when the neural network predicts the reaction of the control object at a certain time interval in the future is selected. Graphic errors of training and testing the neural network as model of a catalytic reactor is given. It is concluded that the control system based on a neural network with the prediction algorithm can be used to control the concentration of the product at the outlet of the catalytic reactor.