Microgrid architectures are typically composed of multiple parallel grid-connected inverters, interconnected via LCL filters to comply with grid code requirements while offering low cost and superior dynamic performance compared to L filters. However, the use of LCL filters in microgrids introduces two types of resonance: extrinsic resonance related to each LCL filter and intrinsic resonance resulting from the coupling effects of multiple parallel LCL filters. These resonances can lead to small oscillations, degrading power quality, or large oscillations, compromising microgrid reliability and stability, especially with variations and poor design of LCL filter parameters. Thus, early detection and correction of these oscillations are essential for resilient microgrid operation. This paper proposes a residual-based adaptive virtual impedance technique to effectively detect and mitigate unwanted resonance frequencies. Unlike previous approaches that primarily focus on static impedance adjustments or non-adaptive methods with slow detection of the undesired resonance frequencies, the proposed method diverges by emphasizing the dual functionality of the virtual impedance. It operates in two steps: first, a residual algorithm swiftly identifies undesired oscillations, producing higher values during instability and low values under normal conditions. Second, an adaptive virtual impedance is introduced to restore stability, with its components continuously computed until the residual value falls below the threshold for normal operation. Unlike previous approaches that often focus solely on harmonic sharing or stability without addressing the rapid detection of oscillations, the proposed technique integrates both detection and mitigation, providing a robust solution for enhancing microgrid stability and reliability in the presence of resonances. A comprehensive case study, including simulations, real-time simulations, and experimental tests, is conducted to validate the technique's effectiveness and superior performance.
Read full abstract