A Model-Based Approach to the Estimation and Control of a Continuously Variable Transmission

Abstract

The Van Doorne continuously variable transmission (CVT) provides a continuous range of gear ratios (GRs) compared to a common discrete transmission, offering a potential improvement in driveline and vehicle efficiency. However, poor understanding of the dynamics of slip inherent to the CVT can lead to conservative control designs unable to realize the benefits. This brief presents a control architecture with the goal of estimating and controlling the CVT using recently developed models, given only sensors likely to be included on a production vehicle. The current state of the transmission and the disturbance torque from the driveline shaft are estimated by an extended Kalman filter (EKF). A model predictive controller (MPC) uses these estimates to optimize a feedback control sequence defined over a receding horizon to improve drivability and efficiency while enforcing safety constraints on the slip. The control scheme is validated through nonlinear simulations of a CVT, which demonstrate the ability to track both GR and slip set points while enforcing constraints.