Artificial neural networks for real-time process control

The purpose of this study was to develop a neural network based on real-time control system for fluid level in a given tank. The tank has an "I" shape with a circular cross sectional area varying across the "I" shaped profile height. The simulated "I" shaped tank has a height of 12 meters and divided into three sections. Each of the sections has a height of four meters. The tank is demarcated into water level marks from O m to 12 m at intervals of 1 each. An ANN is designed and developed for real-time control of water level in the tank. The goal was to enable a smooth change in target level at a constant rate of change inspite of the two nonlinearity points introduced by the "I" shaped structure of the tank, after a certain amount of uncontrolled output flow has occurred.<
>