Design and Implementation of Neuro Controllers for a Two- Tank Interacting Level Process

Abstract: The control of liquid level in tanks and flow between tanks is a basic problem in the process industries. Almost all theprocesses in the industries are non-linear in nature. Designing a controller for a nonlinear process is an important problem. The problem of level control in interacting tank processes are system dynamics and interacting characteristics. In interacting process, dynamics of tank1 affects the dynamics of tank2 and vice versabecause flow rate depends on the difference between the liquid levels (h1 and h2). In this paper, a real-time two-tank interacting level process is taken-up for study. The mathematical model of a twotank interacting process is derived. The hydraulic resistances (R1 and R2) are obtained using Experimental data. The servo and regulatory responses are obtained with conventional PI controller. A neural network based direct inverse and internal model controllers are designed for a two-tank interacting process and its performance is compared with conventional PI controller. To develop model based neuro controllers forward and inverse neuro model are developed, trained and validated. Simulation studies are carried out with direct inverse neuro and internal model neuro controllers for servo and regulatory problems. It is observed that, direct inverse neuro and internal model neuro controllers are giving better results when compared to conventional PI controller. Keywords— Two-tank interacting process, PI controller, Non-linear, Inverse control and Internal model controller.