2019 Volume 13 Issue 2 Pages JAMDSM0037
A nonlinear torque estimator for accurately estimating non-measurable transmission-input-shaft torque in a full-scale automotive vehicle with continuously variable transmission is proposed. The estimator is based on a single-degree-of-freedom, nonlinear drivetrain model; it describes the clearance-induced torsional dynamics of the drivetrain and simulates the non-measurable torque arising from measurable engine-output-shaft torque and drive-shaft torque. Parameters of the estimator are identified from time series obtained from the full-scale vehicle via particle swarm optimization. The estimated transmission-input-shaft torque agrees with the physical torque, suggesting that the estimation quality is above 94.5%. Our estimator extrapolates the transmission-input-shaft torque outside the identified time interval and estimates an unknown sample of the torque measured from a separate trial of the experiment. Our estimator is further improved by delaying the time series of input torques.
