Performance assessment of a neuro self-tuning PI controller to be used at a water treatment plant

Presents the results of a simulation study used as the first step in applying a neuro self-tuning PI controller at a water treatment plant. The training of the adaptive loop, which requires backpropagation of the error through the plant to the controller, is discussed. The benefits of neural control are then demonstrated by comparison with a fixed gain PI strategy for a process whose transfer function structure and parameters change instantaneously at specified time instants