Real-time flow control using neural networks

In this paper, an on-line trained neural network controller is applied to control the flow rate of a process control rig. The neural controller replaces a conventional controller in the forward path. The overall performance of this controller is compared with that of a PID controller in the presence of noise and non-linearity. It is shown that as the non-linearity is added to the system, the PID controller cannot track the set-point changes, however, the neural controller copes well under various conditions.