Control of a pressure tank system using a decoupling control algorithm with a neural network adaptive scheme

The design of a multivariable adaptive decoupling controller and its application to a real-time pressure tank system is presented. The pressure tank is highly coupled and nonlinear. The developed algorithm allows the noise disturbance to be coloured and uses a modification of the Hopfield neural network to identify and track the system parameters. The control algorithm is tested, first in a simulation using an identified model and secondly in a real-time application to a pressure tank system. A very good control performance is reported.