State observation of unknown nonlinear SISO systems based on virtual input estimation

This paper introduces a generic procedure for the state estimation of unknown nonlinear SISO systems, i.e. when no information is available on their structure, possibly time-varying parameters and potential disturbances. This procedure relies on the choice of an arbitrary linear model and the use of a Generic Linear Extended State Observer, whose principle is also introduced in the paper. The proposed approach overcomes well-known model-based nonlinear techniques in the sense that it is easy to implement, all the while avoiding any identification step and mathematical complexity. Simulation results involving a nonlinear system, subject to external disturbances and measurement noise, compare the performance of the proposed approach to the one of some model-free nonlinear observers described in the literature.