Process Control Using a Neural Network Combined with the Conventional PID Controllers

A neural controller for process control is proposed that combines a conventional multi-loop PID controller with a neural network. The concept of target signal based on feedback error is used for on-line learning of the neural network. This controller is applied to distillation column control to illustrate its effectiveness. The result shows that the proposed neural controller can cope well with disturbance, strong interactions, time delays without any prior knowledge of the process.