Frequency domain modelling and identification of nonlinear loads using a least error squares algorithm

Nonlinear electric loads result in harmonics in voltages and current waveforms. This has prompted the need for more accurate models for these loads, in order to allow accurate prediction and mitigation of harmonic distortion. This paper presents a general methodology for modelling electric loads, in the frequency domain, in the presence of harmonics. The proposed method is based on a least squares parameter estimation algorithm and requires advance knowledge of the harmonic contents of load voltage and current waveforms. The residual current, which results from the incomplete extraction of the parameters due to the mismatch between the voltage and the current waveforms, is also calculated. It has been found that the order of the load transfer admittance, as well as the number of its parameters to be identified depend on the number of harmonics common between the voltage and current waveforms. The proposed algorithm is tested using actual recorded data.