Uncertainty identification for a nominal LPV vehicle model based on experimental data

In this paper a practical method is presented for modelling uncertainty of a nominal linear parameter-varying (LPV) vehicle model. The aim with the uncertainty model is to bound the nominal model-error and satisfy robust stability and performance objectives during robust control design. Existing frequency-domain model-validation methods are applied to perform the first aim. The linear fractional uncertainty structure and the distribution of nominal model-error among the uncertainty blocks and disturbances are chosen to perform the second aim. The paper is motivated by the problem of steering a vehicle by alternately braking the front wheels in emergency situations. The identification is performed on real experiment data. The method and the results are demonstrated on a yaw-rate tracking problem and p-controller design on constant scheduling variable of the LPV model. Using the proposed algorithm, on the supposition that nominal model error remains below the bound estimated from the validation data set, an unfalsified model is constructed for robust control guaranteeing robust performance against worst-case uncertainty and disturbance.