Data-Driven Identification of Observed Relative Degrees for Nonlinear Systems

This work presents a data-driven method for automatically identifying the observed relative degrees of general nonlinear systems. The concept of relative degree plays a critical role in many feedback control design techniques for nonlinear systems. However, no general data-driven methods are available for identifying relative degrees from input-output data samples besides techniques optimised for sliding mode control. To address this gap, this work introduces the concept of a signal footprint, which enables the analysis of system responses and detection of the observed relative degree. Validation results demonstrate that the proposed approach is both accurate and efficient in identifying relative degrees from input-output data.