Decoupling internal model control for the robust engagement of clutches

Abstract Research on the robustness of clutch engagement control is attracting considerable interest because of its wide applications in advanced powertrains. Internal model control (IMC) is advantageous because it can handle model errors and external disturbances with light computational loads. However, during clutch engagement, two control inputs (engine output torque and clutch transmitted torque) and two system outputs (driving speed and the driven part of the clutch) are coupled; moreover, the two reference inputs are considerably different. A method for directly decoupling and stabilizing multi-input multi-output systems controlled by a two-degree-of-freedom (2DOF) IMC is proposed based on the construction of V-canonical matrices. The proposed method allows simple expressions for the control algorithm to be obtained without computing an inverse matrix that is used conventionally. Two filters, considering both the reference inputs and system characteristics, are independently designed considering the decoupling operation. The simulation results show that the proposed decoupling 2DOF IMC gains a considerably less tracking error and better robustness to model error and disturbance compared to classical 2DOF IMC and model predictive control. The performance is validated via dynamometer experiments.