Genetic Algorithm Based Parameter Identification of a Nonlinear Full Vehicle Ride Model

Genetic Algorithm is applied to the physical parameter estimation of a full vehicle nonlinear multi-body ride model. Beforehand unity of system representation (identifiability) and sensitivity analysis for determining the effects of parameter changes on the response of the vehicle is discussed. A random road profile is designed as a persistent excitation. Input-output data required for the identification is obtained from ADAMS/CAR simulations of a more complex model. Robustness of the identification method is studied by adding different noise levels to the ADAMS output signals. Validation of the results is carried out by comparison of the identified model outputs with experimental measurements done on the same vehicle, which its ADAMS model was available. Test was performed on the Schenck hydropuls road simulator. Accuracy of estimated parameters is evaluated by information available from other sources such as technical drawings and performance tests of the vehicle parts.