Convergence Filters for Efficient Economic MPC of Non-dissipative Systems

This note presents a novel and efficient Economic Model Predictive Control (EMPC) scheme specifically designed for non-dissipative systems subject to state and input constraints. To address the stability challenge of EMPC for constrained non-dissipative systems, a new concept of convergence filters is introduced. Three alternative convergence filters are designed accordingly to be incorporated into the receding horizon optimization problem of EMPC. To improve online computational efficiency, the variable horizon approach without explicit terminal state constraints is adopted. This design allows for a flexible trade-off among convergence speed, economic performance, and computational burden via simple parameter adjustment. Moreover, sufficient conditions are rigorously derived to guarantee recursive feasibility and stability. The advantages of the proposed EMPC are validated through simulations on a classical non-dissipative continuous stirred-tank reactor.