Constrained Design of Multisine Signals for Frequency-domain Identification of Electric Drive Trains

Abstract

The paper at hand deals with the optimization of multisine signals in terms of effective value for identification of electric drive trains, considering constraints on position, velocity, acceleration and torque. The advantage of maximizing the effective value while respecting the constraints rather than minimizing the crest factor of the input signal is delineated. Results with two algorithms suitable for this optimization task are presented for different sets of dynamic constraints. It is shown that the proposed modified clipping algorithm is much faster than the L1 optimization which is dedicated to simultaneous optimization of several different signals, while being slightly less performant in terms of effective value.