Robust estimation of vehicle sideslip angle from variables measured by ESC system
The vehicle sideslip angle β is one of the most important variables for evaluating vehicle dynamics. In particular, several studies has shown that its knowledge may allow the design of ESC systems having significantly better performances over the standard
- PDF / 439,406 Bytes
- 14 Pages / 419.527 x 595.275 pts Page_size
- 82 Downloads / 145 Views
© Springer Fachmedien Wiesbaden 2015, M. Bargende, H.-C. Reuss, J. Wiedemann (Hrsg.), 15. Internationales Stuttgarter Symposium, Proceedings, DOI 10.1007/978-3-658-08844-6_72
265
Robust estimation of vehicle sideslip angle from variables measured by ESC system
1 Introduction The vehicle sideslip angle ȕ is one of the most important variables for evaluating vehicle dynamics. In particular, several studies has shown that its knowledge may allow the design of ESC systems having significantly better performances over the standard ones based on yaw rate control, see e.g. [1,2]. However, direct measurement of ȕ requires the use of complex and expensive devices (based on optical or inertial+GPS technologies) which cannot be used in production cars. Therefore, the estimation of the sideslip angle from signals measured from standard ESC sensors (steering angle, longitudinal speed, yaw rate, lateral and longitudinal accelerations) has been intensively studied in the last years, see e.g. [3,4,5,6] and the references therein. All these investigations are based on a Two-Step design procedure: a suitable model of vehicle dynamics is first identified making use of experimental data measured on a testing vehicle; then one of existing observer/filtering method (Kalman Filter, Sliding Mode Observer, Moving Horizon Estimator, …) is used for designing the estimator, here called TSVS/SA (Two-Step Virtual Sensor of Sideslip Angle), see Figure 1A.
Figure 1
A: TSVS design procedure
B: DVS design procedure
All these TSVS design procedures suffer from severe drawbacks, which are now briefly recalled. For more documented discussions see [7,8,9,10] and the references therein. Finding optimal estimates (e.g. minimal variance) is computationally intractable when the identified model is nonlinear. Thus, computationally tractable but necessarily approximate methods are then used, whose accuracy is hardly evaluable even if the model exactly describes the vehicle dynamics. However, the filtering algorithms operate on identified models which are only approximate description of the
266
Robust estimation of vehicle sideslip angle from variables measured by ESC system real vehicle dynamic behaviour, whose effects on approximate filtering accuracy is once more hardly evaluable. In addition, the vehicle dynamics may change according to different operational conditions (dry or wet road, tyre wear status, car weight...) whose real-time values are not detected in normal production vehicles. At the best of authors knowledge, no method exists for evaluating the robustness of the TSVS design, i.e if it may give ‘acceptable levels’ of ȕ estimation accuracy despite the problems of approximate modelling, filtering and unknown operational conditions previously discussed. Indeed, it appears that acceptable robustness levels have not yet been achieved from these studies, as documented by the fact that no estimator of sideslip angle designed with these methods has been made available until now on commercial vehicles. In the present paper, a sideslip angle
Data Loading...
