THE PERFORMANCE OF MULTILEVEL PERTURBATION SIGNALS FOR NONLINEAR SYSTEM IDENTIFICATION

摘要

A method for determining the optimal levels of multilevel perturbation signals for nonlinear system identification is described. A performance index for the optimized signals, directly related to the identification accuracy, is based on the condition number of a Vandermonde submatrix of the perturbation signal levels vector. The signal levels are optimized under the constraint that the signal has appropriate harmonic properties for system identification. The performance of all pseudo-random signals with 3, 5 or 7 levels, generated in Galois fields from GF(3) to GF(13), when used for identifying nonlinearities of order up to 6, is determined and compared.