Thermal Clutch Model Observability and Observer Effects of a Torque Sensor in the Powertrain
(English)Manuscript (preprint) (Other academic)
Torque sensors for automotive powertrains are becoming cheaper and a viable option for production vehicles. It is investigated how a torque sensor located on the input shaft of an Automated Manual Transmission (AMT) in a Heavy Duty Truck (HDT) affects the observability of a clutch model for torque transmission that includes thermal effects and aging. In order to handle an offset that is introduced with the new sensor the model is augmented with a random walk and an equation for the engine inertia dynamics. The sensor offset is observable regardless of mode of operation of the clutch and observability of the states in the original model is unchanged. Furthermore the design of an Extended Kalman Filter (EKF) built upon the augmented model requires a new mode of operation compared to an EKF built upon the original model. In addition a mode dependent process noise covariance matrix is used. The new EKF is evaluated using measurement data from an HDT equipped with a magnetoelastic torque sensor. The EKF is demonstrated to work and the estimates converge rapidly, even from poor initial values. It is also computationally simple and therefore suitable for real-time applications.
clutch control, temperature, Extended Kalman Filter (EKF), modeling
IdentifiersURN: urn:nbn:se:liu:diva-108836OAI: oai:DiVA.org:liu-108836DiVA: diva2:733158