摘要：AbstractIn this paper, real-time identification of the dry-clutch torque characteristics model in automotive systems is investigated. The proposed algorithm focused to provide the direct capability to update the clutch torque characteristics model by using engine speed and torque as measurable signals in real-time, within a multiple model predictive control loop. First, the clutch torque is estimated with some uncertainty. Then, it is proposed a new trust-region based identification approach which is robust enough to cover this uncertainty. The aim of the MPC controller is to ensure a comfortable lockup by avoiding engine stall and reasonable engagement time. The effectiveness and performance of the approach are demonstrated through the real-time model simulations in various conditions.
