Model-based estimation of the torque characteristics of an automated dual-clutch transmission in wet technology

In many modern automotive driveline concepts the power transfer between the drive input and output is conducted by a friction clutch. The drivability of a vehicle and the exploitation of the engine power depend primarily on the interaction between the available torque and the torque delivered through the clutch. The clutch characteristic is therefore important in this context. A clutch characteristic defines the relationship between the actuating variable of the contact force and the torque delivered through the clutch. In this paper two different model approaches are discussed for estimating of the clutch characteristics using an adaptive, extended Kalman filter in combination with the recursive prediction error method. The two approaches differ in the level of abstraction and itemization used in the prediction models. The verification studies indicate that both modelling approaches allow an unbiased estimation of the clutch characteristic. However, model-dependent differences arise in transient behaviour (such as convergence time) as well as in the implementation and computational costs.