Comparison of Parameter Identification Methods for Linear Dynamic Systems Under Mixed Noise

We study the possibility of comparing methods with the atypical noise conditions by examing three parameter identification methods and using the sensitivity theory for local expansions of objective functions. The theoretical results are confirmed by numerical experiments on the equations of longitudinal motion of an aircraft, where the parameters are identified by the linear least-squares method, the method of instrumental variables in a frequency domain, and the variational method. The mixed noise (an additive noise in observations and a noise in the residual of the equation) is taken for perturbations.