Online estimation of vehicle driving resistance parameters with recursive least squares and recursive total least squares

We introduce a recursive generalized total least-squares (RGTLS) algorithm with exponential forgetting that is used for estimation of vehicle driving resistance parameters. A vehicle longitudinal dynamics model and available control area network (CAN) signals form appropriate estimator inputs and outputs. In particular, we present parameter estimates for the vehicle mass, two coefficients of rolling resistance, and drag coefficient of one test run on public road. Moreover, we compare the results of the proposed RGTLS estimator with two kinds of recursive least-squares (RLS) estimators. While RGTLS outperforms RLS with simulation data, the recursive least squares with multiple forgetting (RLSmf) estimator provides superior accuracy and sufficient robustness through orthogonal parameter projection with experimental data. On the other hand, RLSmf suffers from serious convergence problems when it was used without parameter projection.