Parameter Optimization for an Accurate Swept-Sine Identification Procedure of Nonlinear Systems

Modeling the nonlinearity of a system is primary importance both for optimizing its design and controlling behavior physical systems operating with wide dynamic range input values, which linearity hypothesis may not be sufficient. To become practical use, identification nonlinear models must accurate computationally efficient. For these reasons, in recent years, among numerous that have been proposed technical literature, Hammerstein model has widely applied as consequence proposal new pattern technique based on pulse compression, makes very applications it adopted. requires characterization linear filters present different branches model. These filters, constitute parameters to identified, considered respect their trends over time or, equivalently, frequency trends, amplitude phase responses. The can if obtained each filter adequately characterize entire specific subjected normal operation identified. This work focuses this aspect, i.e., adequacy identified how obtain correct interest. procedure exponential swept-sine signals defines domain by making use intermediate functions are related impulse responses through transformation. In paper, we analyze, detail, roles bandwidths excitation signal matched filter, basis procedure, verify assumptions made about amplitudes bands, propose criteria defining order maximize accuracy identification. experiment performed possible avoids limitations significantly improves quality results, description domain.