Gain Scheduled Adaptive Model Predictive Controller Design for Quadruple Conical Tank System

Model predictive controllers are widely used in process industries for many decades. This work introduces a scheduling algorithm with Model Predictive Controller (MPC) to control a nonlinear plant, Quadruple Conical Tank System (QCTS). The methodology adopted is to design multiple linear Model Predictive Controllers (MPC) and a scheduler. The controller output will be the weighted aggregate of MPCs, each with a linear state-space model linearized at a specific operating region. Depending on the current level of the plant, the scheduler assigns weight to each controller. A comparative study with MPC is also done and this method has proven to be effective to maintain the desired level over the entire region of operation for both minimum and non-minimum phase characteristics of the plant.