Adaptive low-impact hybrid islanding detection for grid-connected PV inverters in unbalanced and weak power systems

Abstract

Whose fault tolerance, loading pattern, and balance vary over time. In such a feeder, an islanding detection method is expected to operate near the load-generator matching condition, tolerate practical background imbalance, and avoid unnecessary disturbance during normal service. This paper presents an analytical formulation, MATLAB-compatible time-domain numerical simulation, and parametric evaluation of a hybrid islanding detection algorithm for a 50 kW, 400 V, 50 Hz three-phase grid-connected PV inverter controlled by stationary-frame proportional-resonant (PR) current regulators. The proposed method uses a passive detection layer to enable a short negative-sequence reactive micro-probe injection. The injected signature is detected by correlating the measured negative-sequence voltage response against a normalized probe template. The paper defines the passive trigger condition, injected signature, normalized correlation detection metric, adaptive-gain injection procedure, power-quality bounds, latency requirements, and benchmarking configuration. The numerical simulation and parametric results are generated from the stated discrete-time detector model for different inverter ratings, probe amplitudes, gain settings, grid strengths, disturbance inputs, and correlation-response ratios. This evaluation also supports repeatable threshold selection, clarifies the trade-off between detection sensitivity and injected disturbance, and provides a reproducible basis for future EMT, HIL, field, controller, simulation, and multi-inverter validation studies.