A Data-driven Approach to Grid Impedance Identification for Impedance-based Stability Analysis under Different Frequency Ranges

Abstract

The instability caused by inappropriate damping design of grid-connected converters under specific grid impedance makes the grid impedance estimation a crucial issue. To guide the system controller design toward a stable and adaptive system under various operating conditions, a three-stage data-driven approach for grid impedance identification with three different frequency ranges is proposed by taking advantage of massive data coming from measurement and/or simulation. In the case study, Monte-Carlo simulation is adopted for obtaining the grid impedance data under different operating conditions. K-means clustering is used to partition the processed impedance data, and a high order grid impedance model is generated for each frequency range, in accordance with the practice of resonance mitigation design. The estimation results show that with this approach, the grid model in different frequency ranges can be reduced without losing accuracy while having the potential of being more accurate for impedance-based stability analysis.