A Computationally Efficient EIV Models Identification Method Using the V-Gap Metric

A frequency-domain identification method with higher computational efficiency is proposed in this paper. The minimum v-gap metric between the disturbing model and the estimation model is used as an objective function. By change the constraints way for the Nyquist winding condition, the computational efficiency is much higher than the way of previous point-to-point constraint. The proposed method is applicable to not only the errors-in-variables models but also the conventional output noise corrupted models. Parameter optimization of the estimation model employs the interior-point method, and thus the computational efficiency is greatly improved in comparison with the linear matrix inequalities based identification method. The length of experimental data is typically larger than the order of the estimation model, so the calculational efficiency is improved greatly. Finally, a simulation experiment is given to demonstrate that the proposed identification method is computationally efficient.