Automated parameter identification for a dry clutch

Detailed models of automated passenger car transmissions are used for developing new shift/ control algorithms [1], [2]. The modeling process is based on good system knowledge and appropriate parametrization. Without a suitable parametrization the model will not fit the real system behavior. This paper describes a two-step method to identify the mechanical parameters of a dry clutch in order to parametrize a simulation model. This is done at a transmission test bench by using Structured Recurrent Neural Networks and an automated identification process. In the first step the inertias of the clutch are identified. Furthermore the sliding friction of the test bench drive train is also determined. In the second step, the torque capacity of the dry clutch is ascertained.