Rack Force Estimation for Electric Power Steering

This paper discusses a steering rack force estimation scheme using test-rig generated models. In addition to friction identification, a model of the electric power steering system is identified by the use of the instrumented test-rig. It turns out that the friction in the steering system is highly load-dependent, asymmetric with respect to speed, and shows no Stribeck effects. A LuGre model is adopted and fitted to approximate the measured dynamic friction. Consequently, this model is used in a friction compensator which is combined with a linear disturbance observer to estimate the steering rack force. The proposed estimation scheme is analyzed via evaluated system simulations and experiments on the steering system test-rig. Finally, considering the fact that the friction level varies with each steering device manufactured and installed, the paper discusses algorithms for friction level adaptation.