Model Extraction for Power Electronics Systems Using Vector Fitting Based on Sampling Optimized Method

Abstract

Models of power electronics systems are the basis for further analysis and control, so the vector fitting, a widely used data-driven modeling approach, becomes par-ticularly important. However, vector fitting often requires a large number of frequency response measurement points, otherwise it may converge to an inaccurate model. To address this issue, this paper uses the sampling optimized method to obtain a more reasonable distribution of measure-ment points before vector fitting. This paper demonstrates that combining these two methods can effectively improve the accuracy of the fitted model. Numerical tests prove its effectiveness.