State Estimation and Control of Multivariable Interacting Reservoir System

Abstract

This work describes control of a multivariable laboratory process that consists of four interconnected reservoirs. The system can show both, minimum phase and non-minimum phase characteristics at different operating point. The challenge is to regulate the level of bottom tanks to get the desired setpoint inspite of inherent performance limitations due to multivariable right half-plane zeros. Classical controllers are not able to handle multivariable interactions carefully. In this work, we proposed a novel innvoation biased based Linear Quadratic Gaussian (LQG) control scheme in combination with state estimator to achieve efficient servo control and disturbance rejection of quadruple tank process. This approach employs Kalman filter for state estimation and uses estimation error feedback to achieve offset free closed loop behavior. The closed loop responses can be shaped using filters to mitigate sudden changes in the setpoint and unmeasured disturbances. Simulation results are presented for three different cases in this report. Analysis of the proposed simulation results reveals that the proposed Adaptive LQG formulations in combination with Extended Kalman Filter (EKF) is able to track desired trajectory in prescence of disturbances even in non-minimum phase case