Identification of Power Spectra—Linear and Nonlinear Approaches

Linear dynamical filter systems driven by white noise can be used for the generation of stationary and Gaussian processes. A linear approximation method proposed by W. Wedig is presented and generalized for the modelling of homogeneous field processes. In contrast to this, the direct approximation of the power density spectrum results in a nonlinear and - in general - nonconvex optimization problem. However, it can be solved (practically) using a time domain formulation and approximate solutions of the above linear method.