Real-time identification of dry-clutch frictional torque in automated transmissions at launch condition

Abstract

In this work, a new methodology for online identification of the dry-clutch torque characteristics model in vehicular transmission systems is presented. The proposed approach provides a tool with capability to estimate the clutch torque model parameters directly, using the measurable signals of engine torque and speed within a multiple model predictive control loop, in real time. The first step is to calculate the clutch torque which includes some degree of uncertainty. Then, it is presented an advanced identification approach which is based on trust region and has enough robustness to cover the uncertainty of calculated torque. The structure aims to ensuring a comfortable clutch lock-up by keeping reasonable engagement period and avoiding engine stall. Real-time model simulations in various conditions are also presented to illustrate the superior performance and to demonstrate the effectiveness of the algorithm. The identification method has also been tested on real vehicle equipped with 1.3 L turbocharged diesel engine and six-speed automated manual transmission. Different launch maneuvers have been considered to demonstrate the methodology and promising results have been achieved.