Experimental results of model predictive control of a coupled tank system using interior-point barrier algorithm

In model predictive control (MPC), the control input is determined at each instance by formulating and solving an optimisation problem that incorporates real-time measurement updates. Applications characterised by rapid process dynamics often necessitate shorter sampling intervals, which impose stringent requirements on computational consistency. Also, implementing MPC on hardware with constrained computational resources remains a significant challenge. Consequently, the realisation of MPC for such scenarios mandates an appropriate choice of an optimisation algorithm that exhibits consistency in terms of iteration numbers, ensuring optimised solutions within a sampling period in real-time environments. This study introduces the application of an infeasible interior-point barrier algorithm tailored for MPC. The MPC is formulated for linear systems in the condensed framework so that the number of decision variables in the optimisation problem is reduced. The solution algorithm of the optimisation problem is formulated on primal-dual conditions. The proposed algorithm is evaluated in real-time on a coupled tank system. Both the state and output feedback MPC are implemented that demonstrate consistent iteration count throughout its runtime. Comparative performance analyses with other methodologies are conducted elucidating the trade-offs inherent in different approaches.