Higher demands on comfort and efficiency require a continuous improvement of the shift process. During the launch and shift process the clutch control is used to get a smooth and efficient behavior. In this short time of acting the shifting behavior can be rated. Many control concepts use a clutch characteristic to calculate the actuator signal based on the clutch torque. Therefore, a high quality of this characteristic is necessary. Because of the dynamic process during clutch engagement the clutch characteristic needs further information to reach a high accuracy for the control algorithm. In this paper an existing clutch torque characteristic is extended to a characteristic map where the clutch torque becomes a function of the current actuator signal of the clutch and the clutch slip. The extension of the torque characteristic describes the slip based dependencies, e.g. the friction coefficient. The model of the characteristic map consists of the multiplication of two separate functions in these two dimensions. The parameters of this model are estimated using different identification algorithms, in this case a non-linear recursive and a non-recursive identification algorithm. Both estimation algorithms result in a characteristic map of the clutch behavior. A further advantage of the presented approach is the normalization during the estimation process of the slip-based function with the cost function. Thus, the original torque characteristic can be used as fallback if the identification of the slip-based part still need reference data to converge or shows an implausible behavior.
展开▼
