Adaptive Neural Network based control to counter disturbances for Coupled-Tank Water Level control

Abstract

This paper addresses the tough issues pertaining to water level control issues in coupled-tank systems, which might be prevalent specifically to Process industries but has also seen widespread nonlinearities in its behaviour being sensitivity to disturbances of any kind. Here we posed a novel Adaptive Neural Network control-based strategy which excellently compensates parameter uncertainties, for machine nonlinearities, and influx disturbances. This control architecture perfectly combines a conventional Proportional-Integral (PI) controller with a single hidden-layer neural network offering online weight adaptation. Here to approximate and cancel unknown nonlinearities Neural Network based compensator accompany radial basis capabilities has been employed, while the adaption law has been devised Lyapunov Stability theory to insure stability of all closed-loop processes. Extensive MATLAB/Simulink simulations show how well the controller performs in a variety of difficult situations, such as setpoint changes, inflow disruptions, and parameter fluctuations. The proposed controller has some added advantage over PID control exhibited in improvement in Settling Time, decrease in Overshoot , and good disturbance rejection. It also has almost zero steady state error. All these advantageous favours it for real world industrial applications where accuracy matters.