Active Islanding Detection Using Pulse Compression Probing

TL;DR

This paper addresses islanding detection for distributed energy resources by actively probing the grid with Pulse Compression Probing (PCP) and extracting the small-signal impulse response at an inverter terminal. It constructs a state-space model from the impulse response using the Eigensystem Realization Algorithm (ERA) and compares it to a nominal intact-system model via the nu-gap metric $\delta_{\nu}$ to decide islanding, with a practical threshold around 0.9. The method supports simultaneous probing by multiple DERs and is demonstrated on a modified IEEE 34-bus feeder with three grid-tied solar plants, achieving rapid detection within the probing window (up to 223 ms) and strong islanding accuracy in Monte Carlo tests. A physical implementation is shown by embedding the probing signal in inverter switching logic, offering a local, communications-free approach suitable for distribution networks. Limitations include the noiseless simulation environment and a balanced-island scenario left for future work, along with extensions to quantify the non-detection zone and enhance topology detection via multi-model filtering.

Abstract

An islanding detection scheme is developed using pulse compression probing (PCP). A state space system realization is taken from the probing output. The nu-gap metric is applied to compare the measured system to fully intact system and classify it as islanded, or grid-connected. The designed detector displays fast operation, accurate islanding detection results under varying grid condition, and is physically implementable at the terminals of an inverter. The method is verified via electro-magnetic transient (EMT) simulation on a modified IEEE 34 bus test system with randomized loads and simultaneous probing at three independent solar plants, with the probing signal directly implemented into the logic of a switching inverter model.

Figures (4)

• Figure 1: One period of $\sigma(t)$Piaquadio_2023, an impulse-based pseudo-random binary sequence (PRBS) Ljung_1999 of order n = 6, and its corresponding continuous time signal, $p(t)$, a pseudo-random binary pulse train (PRBPT). The PRBPT is used as the probing signal in this work.
• Figure 2: A schematic showing the process of pulse compression probing (PCP) and example signals. A power electronics device injects the probing signal $p(t)$ as a voltage added, $u(t)$, the bus voltage created by other sources in the network. The measured system response (current) is cross correlated with $s(t)$, a repetition of the original probing signal to yield the output.
• Figure 3: A modified version of the IEEE 34 bus test system Kersting_2001. Three distribution-scale, grid-tied solar plants, breakers, and a 5 kW induction motor load are added. Phases are color coded, and breakers are indicated by numbered yellow boxes. PCP is assumed to be available at each solar plant. Probing signals are labeled at Plant 1.
• Figure 4: Implementation of a PRBPT in the firing sequence of a 4-switch H-bridge with neutral. On a rising edge, the first signal (blue) closes switch 2, and opens switch 4, and a delayed signal (red), closes switch 1 and opens switch 3. On a falling edge, the first signal (green) opens switch 1 and closes switch 3, while a delayed signal (purple) closes switch 4 and opens switch 2.