Local identifiability of an LFT structured dynamical system

Rather than just at a specific parameter value, this paper investigates conditions under which a dynamical system with a linear fractional transformation structure is locally identifiable at almost all possible parameter values. Several computationally attractive conditions are obtained, which apply to general systems or some special ones. These conclusions mainly transform system local identifiability into a full-rank property of some matrices. When it comes to generic local identifiability, the corresponding matrices contain several unknown parameters, and it is not straightforward to ascertain their full-rank property. Based on the special structure of the associated matrices, a computationally feasible algorithm is proposed for the verification of system generic local identifiability. Finally, a power system model is adopted as an example to illustrate the validity of the obtained conclusions.