K-gap metric-based switching strategy for MPC with guaranteed cost control

Model predictive control (MPC) has been widely used in industrial processes. However, variations in operating conditions caused by factors such as equipment aging and load changes may lead to mismatch between the operating model and the predictive model and/or controller parameters, thereby degrading closed-loop control performance. To address this issue, this paper proposes a K-gap metric-based supervisory switching strategy for MPC with guaranteed cost control. First, control performance is evaluated using the ISE–TSV index, which combines the Integral Squared Error (ISE) and Total Squared Variation (TSV). Then, when performance degradation is detected, the K-gap between the operating model and the predictive model is used to determine the recovery action. If the K-gap remains within a prescribed range, controller performance is recovered by updating the MPC parameters; otherwise, the predictive model is updated to restore control performance. In this way, the proposed framework coordinates controller retuning and model updating according to the degree of model mismatch, thereby improving performance recovery efficiency while reducing unnecessary computational cost. The stability of the switching process is analyzed theoretically, and the effectiveness of the proposed strategy is illustrated through simulations on a Continuous Stirred Tank Reactor (CSTR) subject to operating-condition changes.