False Data Injection Impact on High RES Power Systems with Centralized Voltage Regulation Architecture

Tommaso Bragatto, Marco Antonio Bucarelli, Maria Sofia Bucarelli, Federico Carere, Alberto Geri, Marco Maccioni

TL;DR

This paper investigates the impact of false data injection attacks on centralized voltage regulation systems in power grids with high renewable energy source (RES) penetration. The authors developed a novel false data generator and a microgenetic algorithm (µGA)-based voltage regulation framework to simulate and analyze these attacks, highlighting the critical need for enhanced cybersecurity in modern power systems.

What Problem Does This Paper Address?

The increasing integration of distributed generation (DG), particularly from renewable sources like solar and wind, is transforming power distribution networks into more complex and dynamic systems. While this shift offers environmental benefits, it also introduces challenges for maintaining voltage stability. DSOs (Distribution System Operators) are implementing centralized voltage regulation (VR) architectures to manage these fluctuations, where DG units adjust their reactive power output based on grid voltage measurements. However, the growing reliance on digital communication and control systems in these networks makes them vulnerable to cyberattacks. False data injection attacks (FDIAs), where attackers manipulate sensor readings, can trick the VR system into making incorrect decisions, potentially leading to voltage violations and power outages. This paper addresses the critical gap in understanding how realistic FDIAs, specifically targeting residential and non-residential customer data, impact these advanced VR systems and the overall grid reliability.

Methods

The authors developed two key components:

  • False Data Generator: This Python-based tool simulates realistic FDIAs. It can generate "spot" anomalies (sudden, localized changes) and "drift" anomalies (gradual shifts in values) or a combination of both. The generator allows for configurable parameters like the percentage of affected users and the scale of the anomaly, aiming to mimic the "capillary penetration" of attacks.
  • Centralized Voltage Regulation Framework: This framework, implemented in Octave using the MATPOWER package, models a centralized VR system. It uses a microgenetic algorithm (µGA) to optimize the reactive power output of PV generators and transformer tap settings to keep grid voltages within acceptable limits (0.95 to 1.05 p.u.). The µGA iteratively adjusts control parameters to minimize voltage violations.

The study then simulates FDIAs by injecting falsified data into this VR framework and observing its performance.

Datasets

  • IEEE 118-bus test system: This standard benchmark distribution network was used as the case study for evaluating the impact of FDIAs.
  • Yearly load profiles: A standard profile from [37] was adapted and scaled for passive loads.
  • PV generator power profiles: An open-access dataset from [38] was utilized for simulating the output of PV generators.

The paper does not introduce a new benchmark dataset but rather uses existing ones to simulate a realistic scenario with high RES penetration.

Key Insights and Ideas

  • Realistic FDIAs Matter: The study emphasizes that simulating FDIAs with realistic data profiles (residential/non-residential) is crucial, as these can have a more nuanced and potentially dangerous impact than manually crafted anomalies.
  • Drift Anomalies are More Disruptive: The research found that "drift" type anomalies, which cause gradual changes in measurements, are generally more disruptive to the voltage regulation system than "spot" anomalies, leading to more frequent and severe overvoltages.
  • Subtle Attacks Can Be Dangerous: A key finding is that a distribution of false profiles causing low measurement variations can be more dangerous than concentrated attacks with large differences from actual measurements. This suggests that attackers might aim for subtle manipulation to evade detection.
  • VR Framework Resilience: Despite the introduction of false data, the developed µGA-based VR framework demonstrated some resilience, successfully avoiding a significant number of voltage violations that would have occurred without regulation. However, its effectiveness is significantly degraded by sophisticated FDIAs.
  • Cybersecurity is Paramount: The paper strongly concludes that the current cybersecurity framework for DSOs is insufficient to handle the potential impact of FDIAs, which can lead to substantial electricity interruptions.

Key Results

  • 25% Node Attack: When 25% of nodes were subjected to mixed spot and drift anomalies, voltage violations (TV1.05) increased from 7.4% (true data) to 31.5%. The maximum voltage reached 1.127 p.u. compared to 1.056 p.u. with true data.
  • Massive Spread (50% & 75% Nodes):
    • With 50% of nodes affected by drift anomalies, TV1.05 reached 64.4%, and TV1.15 rose to 2.3%. The maximum voltage reached 1.181 p.u.
  • With 75% of nodes affected by drift anomalies, TV1.05 was 65.1%, and TV1.15 was 13.3%. The maximum voltage reached 1.215 p.u., significantly exceeding normal operating limits.
  • Drift vs. Spot: Figures 13 and 14 clearly show that drift anomalies consistently lead to higher maximum network voltages and more frequent violations compared to spot anomalies, especially at higher percentages of falsified data.

Limitations and Future Work

  • Limitations:
    • The study assumes that the detection system is unable to detect the injected false data or that no detection system is used.
  • The case study did not consider current limits on branches, assuming they were within limits.
  • The analysis focused on a specific day with maximum reverse power flow.
  • Future Work:
    • The authors propose comparing their results with different voltage regulation mechanisms.
  • They also suggest assessing the resilience of these systems when combined with a false data detection algorithm.
  • Investigating the impact of concentrated vs. distributed attacks further is also a promising avenue.

Summary updated

BibTeX
@inproceedings{bragatto2023false,
  author = {Tommaso Bragatto and Marco Antonio Bucarelli and Maria Sofia Bucarelli and Federico Carere and Alberto Geri and Marco Maccioni},
  title = {{False Data Injection Impact on High RES Power Systems with Centralized Voltage Regulation Architecture}},
  booktitle = {Sensors},
  year = {2023},
  url = {https://openreview.net/forum?id=HH7Cm5xDZ4},
}
Figure