Research problem

Russian hostile actions targeting Ukraine’s critical energy infrastructure have highlighted the urgent need to adequately protect energy systems. Protecting power grids from various threats poses significant challenges due to several main reasons. First and foremost, power grids are vast and highly complex systems, comprising a large number of interconnected components, including power plants, transmission lines, substations, and distribution networks. Additionally, the threat landscape is constantly evolving, with new, more sophisticated threats—both kinetic and cyber—emerging regularly. Beyond cyber threats, unmanned units now represent a major new risk. In the foreseeable future, there will not be sufficient equipment and personnel to protect power plants, substations, and transmission lines from every possible threat. Therefore, optimizing the deployment of available resources is of utmost importance. To address the urgent need for enhancing the resilience of power grids in Ukraine and allied countries, we propose leveraging recent advances in IT technologies, particularly in artificial intelligence (AI) and optimization algorithms. The aim of this research project is to develop R-GRID, a threat prediction simulator that will model the effects of hybrid attacks using military means on power grids and propose optimisation of available protective measures.

Solutiion

Despite the inherent complexity, attacks on and defences of power grids can now be explicitly modelled and analysed. Recent advances in computational power, AI methodologies, and hardware capabilities have made it possible to optimize the protection of even the most complex critical infrastructure systems.

In the R-GRID simulator, we will explicitly model the power grid, available defence resources and strategies, as well as potential attackers and their attack strategies. We will employ various state-of-the-art methods, including the well-known Stackelberg model of conflict between two adversaries—the Attacker and the Defender. In the Stackelberg approach, the Defender determines how defence resources are deployed and how frequently each target is protected. To prevent predictability, the Defender will randomize these actions, such as deciding how often a patrol should inspect power lines in a specific area. The Attacker is assumed to observe how often and with what resources the Defender protects each target. The challenge for the Defender is to optimally allocate resources to minimize the Attacker’s likelihood of success.

Key use cases for R-GRID include:

  1. Identifying the significance of power grid elements in terms of critical infrastructure and socio-economic factors.
  2. Indicating vulnerable points in the network (e.g., due to lack of redundancy) in the context of critical infrastructure and socio-economic factors.
  3. Assessing the significance and vulnerabilities of grid elements in the context of probable attacks.
  4. Identifying and evaluating potential network modifications to enhance resilience.
  5. Utilizing the medium-voltage (MV) network to supply facilities/areas if the high-voltage (HV) network is destroyed or compromised.
  6. Optimizing resource allocation in the event of an evacuation in parts of the country.
  7. Simulating preventive actions using optimization algorithms for defence resource deployment.

FUNCTIONALITIES

Major functionalities of the R-GRID simulator will include:

  1. Importing the user’s power grid data with options for editing.
  2. Analysing the significance of power grid elements based on their operational importance in the context of critical infrastructure and socio-economic factors.
  3. Evaluating vulnerabilities in the grid, such as the grid’s ability to automatically synchronize or supply power to critical elements in the event of failure.
  4. Analysing possible attack scenarios and identifying the elements most likely to be targeted.
  5. Analysing vulnerabilities under complex attack scenarios involving multiple failures.
  6. Simulating the use of the MV network to power facilities/areas bypassing the HV network.
  7. Identifying the best defence measures and strategies, including simulating preventive actions using optimization algorithms.
  8. Simulating resource allocation, including human resources and management centres, in the event of evacuation or optimizing the placement of repair resources.
  9. Modelling potential improvements to the power grid to increase its resilience.

The R-GRID threat prediction simulator will enable targeted investments in detection systems aimed at limiting or delaying hostile actions and enhancing the security of their energy systems. Specifically, the R-GRID tool can analyse solutions to improve the resilience of power grids in allied countries, capitalizing on the ongoing green energy transition, which offers opportunities to restructure and strengthen these systems.