Electromagnetic fields in the air gap of an electric motor produce electromagnetic forces between the rotor and stator. These forces couple the electromagnetic system to the mechanical one. This electromechanical interaction changes the vibration behaviour of the machine, and it may decrease the critical speeds, induce additional damping or cause rotordynamic instability. The experimental validation of theoretical models of these effects requires modal parameter estimation techniques which are reliable and robust. The main aim of this paper is to compare available techniques for the modal parameter estimation. The studied methods were: a) peak picking, b) prediction error, c) polyreference least-squares complex frequency-domain, d) multiple output backward autoregression, and e) polyreference least-squares complex exponential. Experimental data for the comparison was obtained using a standard six-pole 18 kW induction motor equipped with a long flexible shaft. In addition, the theoretical values using a simple electromechanical rotor model were calculated for the test cases. Comparison showed the short-comings of traditional frequency domain techniques and some advantages of the modern frequency- and time-domain techniques.
A Single Step Modal Parameter Estimation Algorithm: Computing Residues from Numerator Matrix Coefficients of Rational Fractions
