Nonlinear Interacting System with Neuro-PID Controller

In this paper, a control methodology for nonlinear interacting system is developed, in which the PID controller is implemented by neural networks trained by mean square error (MSE) criterion. The mean square error can be estimated recursively by utilizing receding horizon window technique. The measurable disturbances in nonlinear interacting system are input to the Neuro-PID controller besides the sequences of tracking error and hence the feed-forward control is combined with feedback control in the developed scheme. The applicability of the developed control scheme is demonstrated on Two Conical Tank Interacting Level System (TCTILS) which exhibits dynamic non linearity and coupling dynamics. Simulation results show that the developed control scheme realized a good dynamic behavior of the TCTILS and a perfect level tracking.