Parameter estimation and validation of a tractor and single axle towed implement dynamic system model

Developing models and estimating model parameters for a tractor and implement system is important for rapid development and improvement of precision implement guidance systems. A dynamic model for a tractor with a towed implement was developed. This model contained a tire lateral force model with several parameters that are varying and difficult to measure accurately. Several field experiments were conducted to collect trajectory data for a tractor with a single axle grain cart. Data from these experiments were used to estimate the tire lateral force model parameters. Three different replicates of the experimental trajectories were collected with each of the step, random and chirp steering inputs. The data was collected at 4.5 m/s forward velocity in two different fields. A two-step optimization process was used to estimate the tire model parameters. First, the experimental data and a set of steady state model equations were used to estimate cornering stiffness parameters. Second, a prediction error minimization method and a dynamic model were used to estimate relaxation length parameters. The parameter estimation process was repeated with each replicate of the experimental data, and the individual estimates were combined using a weighted averaging method. The vehicle model responses with estimated parameters represented the system responses with reasonable accuracy. With the parameters estimated from three different trajectories, the RMSEs for trajectories of tractor and implement CGs varied from 0.05 to 0.83 m. The model-based frequency responses also closely matched with the experimental frequency responses.

► We used trajectory data to estimate soil-tire parameters of a tractor and towed implement model.
► We combined multiple estimates of a parameter to reduce estimation uncertainty.
► Dynamic model with estimated parameters represented system dynamics reasonable accurately.