Real-time identification of vehicle chassis dynamics using a novel reparameterization based on sensitivity invariance

This work presents a novel methodology to identify model parameters using the concept of sensitivity invariance. Within many classical system representations, relationships between Bode parameter sensitivities may exist that are not explicitly accounted for by the formal system model. These relationships, called sensitivity invariances, will explicitly limit the possible parameter variation of the system model to a small subspace of the possible parameter gradients. By constraining the parameter identification or adaptation to a model structure with uncoupled parameter sensitivities, a more efficient identification can be obtained at a reduced computational and modelling cost. As illustration, an identification method of using sensitivity invariance is demonstrated on an experimental problem to identify, in real time, a time-varying tire parameter associated with the chassis dynamics of passenger vehicles at highway speeds. The results are validated with simulations as well as an experimental implementation on a research vehicle driven under changing road conditions.