In English

Realtime estimation of tyre-road friction for vehicle state estimator

Jakub Prokes
Göteborg : Chalmers tekniska högskola, 2015. Diploma work - Department of Applied Mechanics, Chalmers University of Technology, Göteborg, Sweden, ISSN 1652-8557; 2015:11, 2015.
[Examensarbete på avancerad nivå]

Tyre-road interaction is the major source of forces by which the road vehicle can be controlled in its speed and direction. These forces are generated by either steering the wheels (lateral tyre force), applying driving or braking torque on wheels (longitudinal tyre force), or both. These forces are limited in magnitude and crossing this limit can in the worst case lead to the loss of control. Therefore it is of highest interest to keep the tyre forces below the limit. The limit is commonly denoted as the tyre-road friction coefficient, and it relates the maximal achievable resulting horizontal tyre force (lateral, longitudinal or their combination) with the actual vertical load on the respective tyre. Knowledge of the tyre-road friction coefficient would thus allow to limit the requested control forces so that the loss of control is avoided. The goal is thus to estimate the tyre-road friction coefficient in realtime during operation of the vehicle. In this thesis some of the possible ways to estimate the friction coefficient, that has been published earlier, are reviewed. In this thesis the possible methods are limited to those that use in-vehicle measurements and observation of the dynamic behaviour of vehicle and its tyres. Models of dynamic behaviour of tyre to either lateral or longitudinal load is in particular used to identify the friction coefficient. As a model to match the real tyre response a brush tyre model is used and no a priori knowledge of tyre parameters is used. The operation of the vehicle is described as a state machine of specific driving situations, hereafter called driving modes. In this case the vehicle under investigation is the Volvo FMX rigid truck in 8x4 configuration. Cornering and pure longitudinal acceleration driving modes have been briefly evaluated using a simulation model for possibility of friction estimation. The pure longitudinal acceleration driving mode proved to be more viable option at the time, especially due to rollover tendency of trucks. A recursive estimation algorithm based on an Extended Kalman Filter (EKF) was developed. Tuning of the EKF is discussed with respect to neccessity of sufficient excitation of the tyre in order to extract information about the friction. The performance of the estimator was evaluated in a simulation model of the whole vehicle with added noise on the input signals to the estimator. In these conditions the estimator is able to successfully predict the tyre-road friction coefficient with sufficient tyre excitation. Experimental tests were performed using the investigated vehicle to gather real input data for the estimator. The processing of the real vehicle measurements is discussed with respect to the estimator functionality. The real data does not match the expected tyre response characteristics which prevent the estimator from predicting the tyre-road friction coefficient. The estimator can however still indicate the low tyre-road friction within a short time interval after saturation of the tyres. Testing of the estimator implemented as a realtime solution in the embedded hardware has not been done due to the issues with in-vehicle measurements and signal processing. The uncertainty of the inputs to the estimator based on experimental data has been evaluated and a sensitivity of the inputs has been analysed. The sensitivity of the estimator to the excitation conditions is discussed and conditions for when the estimator can update the estimates are introduced. A process to identify the sufficient excitation is developed and tested in the simulation model. This process can serve as a base for adaptive tuning of the EKF estimator. The functionality of this process has not been evaluated in the conditions of real inputs.

Nyckelord: tyre-road friction, estimation, realtime



Publikationen registrerades 2015-09-25. Den ändrades senast 2015-09-25

CPL ID: 223197

Detta är en tjänst från Chalmers bibliotek